Professional Documents
Culture Documents
General Editor
Margaret Levi University of Washington, Seattle
Associate Editors
Robert H. Bates Harvard University
Torben Iversen Harvard University
Stathis Kalyvas Yale University
Peter Lange Duke University
Helen Milner Princeton University
Frances Rosenbluth Yale University
Susan Stokes Yale University
Sidney Tarrow Cornell University
Kathleen Thelen Northwestern University
Erik Wibbels Duke University
JOHN M. CAREY
Dartmouth College
CAMBRIDGE UNIVERSITY PRESS
Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore, São Paulo
www.cambridge.org
Information on this title: www.cambridge.org/9780521884938
© John M. Carey 2009
Preface page ix
vii
Contents
viii
Preface
Shortly after the 2006 election, in which the Democrats recaptured control
of the U.S. Congress, the spoof newspaper The Onion ran a story in which
Nancy Pelosi, the new Speaker of the House, reprimanded her partisan
colleagues for supporting her legislative agenda without necessarily meaning
it. Referring to a fictitious bill, The Onion had Pelosi admonishing Democrats
not to ‘‘just pass it because I want it, but because you want it, too,’’ and went
on to describe Pelosi’s ‘‘concern that her relationship to the House was based
completely on voting’’ (The Onion, 42 [49], December 4, 2006).
Legislative decisions are about votes, and voting behavior is organized
by parties. If we want to understand legislatures and the representation they
provide, it makes sense to look at partisan voting. To The Onion, the joke
was that Pelosi might care about anything beyond that bottom line.
It never got big laughs, but I had a similar idea in mind around a decade
ago, when I started the project that became this book. At the time, the study
of voting in the U.S. Congress was a bustling cottage industry, but there
was almost no information about legislative voting outside the United
States. The reason, it seemed to me, had to be the lack of available data
on votes. So, to begin, I set out to collect data on votes in a number of
legislatures, mostly in Latin America where I had some experience, but also
in other assemblies where I could establish research connections. My first
surprise was that, in most countries, it was exceedingly unusual to record
how each legislator voted on a given proposal. What The Onion took to be
the bedrock of legislative representation could not be taken for granted in
many democracies.
As I explored the issue across more and more assemblies, it became clear
that a prior question – before how legislators vote – is whether assemblies
make it possible to know how legislators vote. So the research agenda
ix
Preface
evolved and expanded, and I spent as much time talking with politicians,
journalists, and activists about whether they favored voting transparency,
and why, as I did collecting and analyzing voting data.
As it turns out, I spent a lot of time on each, which accounts for the ten
years that passed between starting the project and publishing this book.
Those years have seen progress in the study of legislative voting beyond the
halls of the U.S. Congress. This book takes a step toward mapping, and
explaining, the world of partisan voting in legislatures. Data availability
remains an obstacle in most assemblies. Many still record few or no votes,
and those that do record often do not make vote records easy for outsiders
like scholars, or citizens, to examine. The problem is more than academic.
Lack of voting transparency is also an obstacle to accountability.
There is much more on this topic in the book itself. Here, I want to
recognize and thank the organizations and the people who made my re-
search possible. The book offers the broadest cross-national analysis of
recorded voting to date, and all the data collected for the project are avail-
able online for other researchers to use. Doing field research in ten coun-
tries, and collecting the data from fifteen others, required resources,
expertise, and effort beyond what I could muster on my own. Early
financial support was provided by National Science Foundation Grant
SES-9986219 and also by the Weidenbaum Center on the Economy,
Government, and Public Policy at Washington University in St. Louis.
I received outstanding research assistance at Washington University
from Christopher Kam, Connor Raso, Meg Rincker, John Bunyan, Sarit
Smila, Alba Ponce de León, Erica Townsend Bell, Gina Reinhardt
Yannitell, Adam Bookman, Rachel Kaul, Cheryl Boudreau, Juan Gabriel
Gómez Abellardo, Amy Nunn, and (now Senator) Jeff Smith. Jeff Staton’s
contributions are better described as collaboration than as research assis-
tance, and I continue to learn from Jeff. Rebecca Cantú provided solid
assistance during my brief visit at Harvard. At Dartmouth, assistance from
Anne Bellows, Justin Brownstone, Xavier Engle, and Seth Goldberg helped
bring the project across the finish line.
In the course of conducting field research and in collecting data from
assemblies far and wide, I drew on the expertise, and often on the hospitality,
of dozens of generous souls. Eduardo Alemán, Mark Jones, Valeria Palanza,
Roberto Sabá, and Mariano Tommasi shared data and provided insights into
Argentine politics. In Bolivia, thanks go to Diego Ayó, Carlos Cordero,
William Culver, René Mayorga, José Rivera Eterovic, and Eduardo Rodri-
guez. On Brazil, I am grateful to Barry Ames, Octavio Amorim Neto, Scott
x
Preface
xi
Preface
xii
1
1.1. Introduction
1.1.1. Overview
Legislatures are, formally, the principal policymaking institutions in mod-
ern democracies. The most fundamental policy decisions – budgets; trea-
ties and trade agreements; economic, environmental, and social regulation;
elaboration of individual and collective rights – all must be approved by
legislatures. What forces drive legislators’ decisions? What different polit-
ical actors place demands on legislators, and how do legislators’ actions
reflect these demands?
These are questions about what sort of representation citizens can expect
from those they send off to deliberate and make policy decisions on their
behalf. Citizens want legislatures to be decisive – that is, to resolve the issues
before them without chronic deadlock. They also want accountability, which
entails responsiveness on the part of legislators to citizens’ demands. In mod-
ern democratic legislatures, the principal vehicles for delivering decisiveness
are strong political parties. Decisiveness through party discipline, however,
presents a dilemma in terms of what kind of accountability is possible.
This book distinguishes between collective accountability and account-
ability that operates at the level of individual legislators, which often make
different demands on legislators. In modern democratic legislatures,
collective accountability operates primarily through parties and requires
legislators bearing a common party label to act in concert. Individual
accountability implies a more direct link between a legislator and citizens
and may require the legislator to act independently of party demands.
Individual accountability also requires that information about each
1
To Whom Are Legislators Accountable?
2
Introduction
legislators and how their relative influence affects legislative party unity.
This account, dubbed the competing principals model, generates hypoth-
eses tested against voting data from legislatures in nineteen countries. By
documenting what information is available about legislative votes and pro-
viding new tools to process the information, the book outlines the mix of
collective and individual accountability that legislators deliver across an array
of countries, as well as the potential for political reforms to alter that mix.
The rest of this chapter establishes vocabulary and concepts on which
the book depends. After defining some key terms used throughout the
book, I describe the unique role of political parties in organizing legislative
processes and as intermediaries of accountability between citizens and their
representatives. Then I contrast the ideals of collective and individual ac-
countability and discuss how electoral rules shape the balance between
collective and individual representation. Finally, I present the competing
principals model of demands on legislators and outline the plan of the
chapters that follow.
1.1.2. Definitions
1.1.2.1. Accountability. The expectation of accountability implies
a relationship between a legislator and some other actor or actors
(principals). Accountability means that legislators are responsive to the
preferences and demands of their principal(s), that information about
legislators’ actions is available to the principal(s), and that principals can
punish legislators for lack of responsiveness.
Accountability depends on professional ambition among legislators.
Professional ambition may be a purely venal desire for personal advance-
ment, or a purely altruistic desire to serve others by promoting policies that
advance some conception of the public interest, or some combination of
these. Whatever the motivation, ambition implies the desire to cultivate
electoral resources – renomination, or else nomination or appointment to
an even better office, campaign financing, and good favor among voters. It
also implies that legislators value access to resources within the legislature
itself, such as leadership positions, assignments to key committees, access
to support staff, big offices, perks, and such. Ambitious legislators curry
favor with political actors who can provide these key resources. The ability
to withdraw favor, and so deny the resources that fuel professional advance-
ment, is the enforcement mechanism behind accountability. Overall, ac-
countability should maximize legislative effort and responsiveness to the
3
To Whom Are Legislators Accountable?
4
Decisiveness Problems
5
To Whom Are Legislators Accountable?
formal offices that control action within the legislature. Moreover, Carroll,
Cox, and Pachon (2006) demonstrate that, as democracies mature, parties
expand their control over the offices that determine legislative activity, and
the distribution of these offices among parties grows increasingly regular.
In short, as party systems stabilize, so do the key partisan elements of
legislative organization.
6
Collective versus Individual Accountability
1
Levels and sources of party unity have been extensively examined in the U.S. Congress,
where scholarship has been preoccupied for more than a decade with parsing to what extent
levels of party voting are due to like-mindedness among copartisans (cohesiveness) versus
pressure from party leaders (discipline) (Krehbiel 1998; Cox and Poole 2004). For ‘‘un-
mapped’’ legislatures, the basic question of how unified parties are is of prior concern to the
cohesiveness-versus-discipline matter. Whatever its cause, some measure of voting unity is
necessary for party labels to convey information that is useful to voters. I take up the matter
of overall level of voting unity in Chapter 5 and the issue of cohesiveness versus discipline in
Chapter 6.
7
To Whom Are Legislators Accountable?
8
Collective versus Individual Accountability
2
The brief discussion that follows here of Iraq and Afghanistan is not meant to serve as
a thorough review of legislative electoral rules, much less as a comprehensive analysis of
the politics of these countries. The former is provided in an impressive literature on com-
parative electoral systems (Duverger 1954; Taagepera and Shugart 1989; Lijphart 1994;
Cox 1997), and the latter is well beyond my capacity.
9
To Whom Are Legislators Accountable?
3
One compelling motivation for this choice had to do simply with logistics of electoral
administration: Iraq lacked a reliable census by which legislative seats might be apportioned
across districts according to population.
10
Collective versus Individual Accountability
11
To Whom Are Legislators Accountable?
conflict of interests between the party and its individual politicians.4 Parties
seek to win as many seats as possible. Individual politicians may prefer to be
members of strong parties, but their first priority is to win office. Under
SNTV, candidates who seek to minimize the risk of individual defeat have
incentives to draw votes away from copartisans, undermining the collective
goal of translating votes to legislative representation efficiently. By
privileging electoral individualism, SNTV presents formidable challenges
to parties’ ability to foster internal cooperation among politicians and thus
to provide collective representation (McCubbins and Rosenbluth 1995;
Cox and Thies 1998; Marlowe 2007).
An even more immediate challenge to the feasibility of SNTV in Afgha-
nistan was the incompatibility between individualistic legislative represen-
tation and the representation of women. The Afghan Constitution requires
that at least two lower-house legislators from each of the country’s thirty-
four provinces be female (Art. 83). SNTV provides no alternative basis
other than individual vote totals for awarding legislative seats, so unless
at least two of the top candidates in each province were women, the Afghan
system requires bypassing male candidates with more votes in order to seat
female candidates with fewer votes. In a society where gender-based
inequalities in personal resources, as well as gender bias among voters,
constrain the viability of female candidates, this outcome was inevitable.
In the September 2005 election, 19 women were elected on the basis of
their vote totals alone, but 49 additional women were awarded seats in the
Loya Jirga despite having won fewer votes than 422 other male candidates
(Reynolds 2006). In sharp contrast to Iraqi gender quotas, the purely per-
sonalized and individualized character of legislative voting in Afghanistan
throws into stark relief the mechanism by which quotas delivered these
women to their seats while male candidates with higher vote totals lost.
The fundamental contrast in the Iraqi and Afghan choices over electoral
rules, at this point, is between privileging collective and individualistic
representation. For myriad reasons, the system chosen in Iraq leans toward
the former. This facilitated the initial, descriptive representation of various
collective identities – most notably by party alliance, ethnicity, religion,
4
The problem is also increasingly severe as district magnitude increases. Magnitudes in
Japanese SNTV elections ranged from three to five. In Afghanistan, the median district
magnitude for parliamentary elections was seven, with a third of districts electing ten
or more representatives, and the largest, Kabul, electing thirty-three (Constitution of
Afghanistan, Article 82; Reynolds 2006).
12
Collective versus Individual Accountability
13
To Whom Are Legislators Accountable?
14
Legislators, Principals, and the Structure of Accountability
15
To Whom Are Legislators Accountable?
VOTERS
PARTY
LEGISLATOR
Figure 1.1 Party-dominant representation.
5
In Chapter 6, I also consider broader legislative coalitions – government versus opposition,
for example – but parties are the component units of such coalitions, so this is simply
a matter of moving to a higher level of aggregation.
16
Legislators, Principals, and the Structure of Accountability
Next, consider a political system where voters have the ability to reward
and punish individual legislators directly, perhaps because primary elections
determine nominations, or because party lists are open and candidates win
legislative seats according to their individual preference votes, or because
there are no party lists at all and multiple candidates from each party run
either in a free-for-all format or under a transferable vote rule.6 Under any
of these circumstances, the voters are a legislator’s direct principal. Because
party labels are generally attached to the candidates for whom voters vote,
and because in the aggregate a party’s fortunes depend on the success of its
candidates, voters are also indirectly principals to the parties. Meanwhile,
party leaders, in all likelihood, remain important principals for legislators,
to the extent that they control resources within the assembly itself that
legislators value. They may also retain control over electoral resources, such
as influence over nominations and financing. Thus, legislators now confront
two principals, who may well make competing demands (see Figure 1.2).
We could add a directly elected president to either of these scenarios, as
in Figure 1.3. The formal powers of presidents over the legislative process
vary enormously, but most control access to coveted appointed posts, and
many are endowed with authority to shape the legislative agenda directly,
to veto all or parts of bills, and to offer counterproposals to legislative
initiatives (Shugart and Carey 1992; Aleman and Tsebelis 2005). The array
of powers of most directly elected presidents provides substantial leverage
with which to influence legislative behavior. As with a direct electoral
VOTERS
PARTY
LEGISLATOR
Figure 1.2 Parties and voters as competing principals: Individual-predominant
representation.
6
See Carey and Shugart (1995) for more details on the variety of electoral systems and their
relationship to the relevance of the personal vote, as well as for empirical examples.
17
To Whom Are Legislators Accountable?
VOTERS
PARTY PRESIDENT
LEGISLATOR
VOTERS
PARTY PRESIDENT
LEGISLATOR
Figure 1.3 Presidents as competing principals with party leaders.
18
Legislators, Principals, and the Structure of Accountability
Brazil
VOTERS
LEGISLATOR
VOTERS
EU LEGISLATOR
Figure 1.4 Increasingly complex sets of principals.
their countries of origin. Figure 1.4 suggests how such additional principals
might map onto the accountability relationships described so far.
As the number of principals grows, the potential for competing pres-
sures to pull legislators away from solidarity with their party increases, and
we should expect party unity to drop.
Does the lack of party unity indicate the existence of legislative in-
dividualism? I interpret it as such, for a couple of reasons. One is that
deviance from the party line often indicates an effort by a legislator to act
on behalf of a constituency of citizen supporters, independently from any
mediating forces. This story is associated with the idea of the personal
vote and of individual-level accountability in the simplest sense (Cain,
Ferejohn, and Fiorina 1981; Carey and Shugart 1995). But what if a leg-
islator is pulled away from the national party by another institutional
19
To Whom Are Legislators Accountable?
political actor – by the president, for example? Then deviance from party
loyalty might not be regarded as simply a matter of individual volition.
Yet, as Sophocles’ Antigone discovered, cross-pressures, by their nature,
turn the decisions individuals make into acts of self-definition. In this
light, a legislator who lines up with the president (or governor, or whom-
ever) in contradiction to her legislative party leadership is staking out
a position, even if reluctantly, for which she can be held individually
accountable. Finally, although the number of potential principals making
demands is not limited, legislative decisions almost always come down to
binary choices – voting aye or nay. We may sometimes know what side
a president is on, or a governor, or an interest group. Less reliably still can
we draw any inferences about what the legislator’s constituency support-
ers demand. We can know, however, what side a legislator’s party is on by
how her copartisans voted.
In short, party unity is the point of reference because legislatures every-
where are partisan, because legislators everywhere answer to party leaders
as principals, and because party unity can be identified by the fundamental
act of legislative decision making – that is, voting. Where party unity is
lacking, there are various stories we can tell about legislative motivation,
but they all involve legislators’ making decisions to deviate from the game
plan of the team – the collective unit – that is the central basis of legislative
organization. It is no accident that the term ‘‘party line’’ has entered our
vernacular to connote the antithesis of individualism and independent
thinking. I contrast party unity and collective representation, then, with
disunity, which I associate with legislative individualism. The type of ac-
countability that is possible in any given legislature depends on what sort of
representation is provided.
20
Plan of the Book
21
To Whom Are Legislators Accountable?
available for most legislatures, and the records that are available are gen-
erally light on context beyond the digital ones and zeros of aye and nay
votes. Chapter 5 describes the recorded vote data, as well as the various
indices I use to turn the vast matrices of legislators’ aye and nay votes into
statistics that describe either voting unity at the party or coalition level or
the alternative, legislative individualism. That chapter also presents statis-
tics that describe the levels of party unity in the various political systems for
which I have collected recorded votes.
Chapter 6 develops and estimates a statistical model of the factors that
affect the diversity of principals to which legislators are accountable. The
analysis shows that party unity is highest when party leaders are the dom-
inant principal – that is, in parliamentary systems where voters do not cast
personal preference votes among candidates. Factors that subject legisla-
tors to influences from additional principals can diminish voting unity.
Electoral rules that provide for a personal vote among copartisans have
this effect. The presence of an independently elected president can also
reduce unity, particularly among legislators allied within the president’s
own party and governing coalition. Presidents, in short, are as much lia-
bilities as assets to their legislative allies.
The final chapter concludes with observations about the distinction
between legislative individualism and individual accountability, and the
potential for compatibility between the latter and collective accountability.
22
2
As for party cohesion in Congress, the parties have done little to build up the kind of
unity within the congressional party that is now so widely desired. Traditionally
congressional candidates are treated as if they were the orphans of the political
system, with no truly adequate party mechanism available for the conduct of their
campaigns. Enjoying remarkably little national or local party support, congressio-
nal candidates have mostly been left to cope with the political hazards of their
occupation on their own account. A basis for party cohesion in Congress will be
established as soon as the parties interest themselves sufficiently in their congres-
sional candidate to set up strong and active campaign organizations in the constit-
uencies. (1950:21–22)
23
Collective Accountability and Its Discontents
In short, strong parties have long been held in high academic esteem. In
the next section, I describe how legislative parties in a variety of Latin
American legislatures where I have conducted research reach and enforce
collective decisions. This is a description of the mechanics by which the
collective vision of representation through parties might be realized. Po-
litical reformers often see things differently. It is not that they aspire to
feckless parties. Nor, indeed, would most academics who call for strong
parties aspire to Leninist centralism. But whereas academic observers have
been inclined, on the whole, to see parties as weaker than they ought to be
and needing fortification, the general tide of reform in Latin America has
24
Legislative Parties and Discipline in Latin America
run against the authority of central party leaders, in the name of increasing
the accountability of individual legislators.
1
Generally, the connection between electoral and legislative parties is straightforward, but it
may not be. Party switching in the legislature between elections is common in some coun-
tries, particularly where legislators are elected on the basis of personal votes, where volatility
in party support is high, or both (Thames 2007). Brazil is notorious on both counts (Despo-
sato 2006b; Samuels 2000). Rules of procedure in many legislatures also require some
minimum membership level for registration of a party group, so parties that have insuffi-
cient numbers may be forced either to coalesce in the legislature or to forgo whatever
resources are allocated to groups.
2
Venezuela’s 1999 constitutional provision (Art. 201) prohibiting such constraints is unusual
in this respect.
25
Collective Accountability and Its Discontents
from, or imposed by, the national party organization, the norm is that
decisions are made in party group meetings by majority rule. This applies
to the question of whether to require discipline (or, alternatively, to leave
a matter open to conscience) and, if so, what the party line should be.
In a few cases, a provisional position for a party group can be set by the
group’s leaders themselves, although in cases where such an initiative
prompts dissent the fallback is to deliberate and vote within the party
group. According to Hugo Carvajal, Movimiento Izquierdista Revolucio-
nario deputy and former president of Bolivia’s Chamber of Deputies,
The bancada decides [its position] depending on the parliamentary rhythm; the
parliament has rhythms. The consultation sometimes gets only as far as a bancada
leader, and he defines a position and then transmits it to the group – we could say he
‘‘socializes’’ with the members – this decision that he’s already made and has adop-
ted in the name of the collegiate body. Sometimes this produces short circuits in the
members’ reaction.
In the case of such short circuits, the remedy is deliberation and a vote
within the bancada.
2.2.2. Discipline
Across the overwhelming majority of parties in the countries where I
conducted interviews, legislators reported that most votes are matters of
discipline. Without estimating precise rates of discipline, Salvadorans con-
curred that open votes are rare events (A. Alvarenga and Duch interviews).
Similarly, Carvajal estimated party-line voting 85 to 90 percent of the time
in Bolivia’s MIR, Guillermo Landazuri estimated the rate to be 90 percent
within Ecuador’s Izquierda Democratica, and Xavier Neira reckoned it
higher within that country’s Social Christian Party (Carvajal, Landazuri,
and Neira interviews). In Costa Rica, open votes appear to be slightly
more common, as do breaches of the party line on disciplined votes. The
chief of staff to the Costa Rican Assembly’s mesa directiva, Eladio Gonzalez
(interview), estimated that across all parties 80 to 85 percent of votes are
subject to discipline among all parties.
The most noteworthy exception to regular decision making at the level
of the party group was in Colombia, where legislators from both major
parties, the Liberals and Conservatives, in both chambers reported that
there are infrequent group meetings and that the majority of votes are left
open, without any established party decision or direction on how to vote
26
Legislative Parties and Discipline in Latin America
3
Colombia subsequently reformed the extremely individualistic electoral system to which
Andrade refers in this comment, replacing it with one that retains substantial individualism
but increases somewhat the extent to which copartisans’ electoral fates are intertwined.
27
Collective Accountability and Its Discontents
28
Partisan Representation Falling Short
did not accept party decisions could not aspire to electoral posts. Everyone knew I
wanted to run for mayor of Managua, and this way I couldn’t be nominated. It’s
almost certain that they won’t permit me to run for reelection as a deputy either.
And they took other measures. I was vice president of the Assembly’s executive
committee, and they took that away, and they won’t let me chair any committees.
29
Collective Accountability and Its Discontents
30
Partisan Representation Falling Short
4
Ecuador, meanwhile, combines personal voting in two-seat districts with closed-list pro-
portional representation in an upper tier.
5
Mexico used a straight SMD plurality until the 1970s, approaching the mixed system from
the opposite direction, adding PR seats gradually from the late 1970s to the early 1990s, to
allow for minority-party representation while maintaining the advantage that SMD plural-
ity tends to provide for the largest party (Molinar Horcacitas and Weldon 2001).
31
Collective Accountability and Its Discontents
The same motivation spurred the shift from pure closed-list PR election
to SMD-PR in Bolivia in 1994 where the plummeting stature of political
parties, evident in street protests as well as opinion polls, was understood as
a demand from voters ‘‘that deputies should be known and acknowledged
representatives of their constituencies and not anonymous representatives
of party leaders. Direct connections between deputies and voters would
therefore enhance the legitimacy and representativeness of the parliament,
making possible the responsiveness and accountability of deputies to their
constituencies’’ (Mayorga 2001). Precisely the same arguments were
made by advocates of a proposal for SMD-PR in Costa Rica in 2000
(Sibaja interview).
Whether these reforms deliver enhanced individualistic representation
and accountability remains an open question. In Bolivia, an early review of
the local responsiveness of SMD deputies (called uninominales) found only
modest improvements (Culver and Ferrufino 2000). In Venezuela, elec-
toral reform in the late 1980s proved insufficient to salvage public support
for the traditional parties and avert their complete collapse in the 1990s.
Yet the COPRE’s recommended SMD-PR format survived even President
Chavez’s constitutional overhaul of 1999; indeed, the champions of Cha-
vez’s reforms echoed the COPRE’s calls for strengthened electoral ties to
local constituencies (Combellas, Tarek Saab, and Fernández interviews).
Finally, although the empirical focus of this book is primarily Latin
America, it is worth noting that the arguments made in that region resonate
as well among SMD-PR advocates in Europe and elsewhere. In the past
decade and a half, Italy, Japan, New Zealand, the Philippines, Russia, and
Ukraine have adopted mixed SMD-PR electoral systems.6 Richard Katz
describes the Italian electoral reform of 1994 as motivated by popular
demands for alternation in government and for ‘‘direct accountability of
individual members of parliament to their electors. There was a desire to
free the electorate from the confines of party labels and ideologies, and to
allow electors to take into account the character, qualifications, and per-
formance in office of individual candidates when casting their votes’’ (Katz
1994:103). Historically, the ‘‘mother of mixed systems,’’ in Germany, was
distinguished by its advocates from the prewar Weimar system of closed-
list PR by its virtue of strengthening the connection between voters and
individual representatives. As noted by Susan Scarrow (2001b:63),
6
Germany has used such a system since 1949.
32
Partisan Representation Falling Short
German advocates of mixed-member rules argued that such rules would ‘‘person-
alize’’ voters’ choices by letting them choose individual representatives from small
districts – indeed, Germans still refer to their system as being an example of ‘‘per-
sonalized PR,’’ a label that is meant to distinguish it from proportional systems that
lack a nominal tier.
33
Collective Accountability and Its Discontents
34
Partisan Representation Falling Short
7
Another 11 percent each said it depended on the circumstances, or offered no opinion.
8
The menu is, admittedly, a bit ambiguous, in that the details of selection within these
groups were not spelled out.
9
The plurality choice was by citizen groups (34%), with other responses including munic-
ipal committees (11%), indigenous groups (6%), unions (5%), and no opinion (15%).
10
The Dominican Republic previously used closed-list PR. Colombian lists were also closed,
although each party was allowed to present multiple lists, injecting substantial individual-
ism into Colombian elections.
35
Collective Accountability and Its Discontents
36
The View from the Chamber
Figure 2.1 How much do you take the opinion of national party leaders into
consideration when making political decisions?
37
Collective Accountability and Its Discontents
Figure 2.2 How much do you take the opinion of voters in your district into
consideration when making political decisions?
38
The View from the Chamber
Figure 2.3 How much do you take the opinion of the government into
consideration when making political decisions?
39
Collective Accountability and Its Discontents
Figure 2.4 Do you think the national party leadership should have more power
over legislators or less?
numbers favor decreasing party leader control overall. In every country but
Colombia, more respondents said ‘‘less’’ than ‘‘more,’’ generally many times
more. In ten of fifteen countries, an outright majority of respondents in-
dicated a preference for less central party control. Across countries, the mean
level of support for increased party control is 13 percent, whereas the mean
support for decreased control is 56 percent.
According to the surveys, legislators claim they prefer more of their own
discretion, and less control from their parties, toward the expressed priority
of representing the interests of voters from their districts. All this may be
posturing, of course, if legislators for some reason felt obliged to dissemble
on the surveys. Even if the relative commitment professed for district voters
versus party leaders or presidents is not sincere, it indicates a professed
commitment to the sort of representation generally associated with legis-
lative individualism rather than collectivism.
40
The Shift toward Individual Accountability
41
Collective Accountability and Its Discontents
42
3
Let us make public the names of those who voted in favor, so our children will know
whom they should curse.
Russian legislator Yuli Rybakov, on a proposal to accept nuclear waste from other countries in
exchange for cash (National Public Radio 2001)
43
The Supply of Visible Votes
The broad question motivating this chapter, and the next, is whether the
information necessary to make individual accountability possible is available
from legislatures. Why tie the transparency of information about legislative
votes so tightly to individual but not collective accountability? This chapter
argues that the responsiveness of individual legislators depends heavily on
which of their potential principals can easily monitor their actions. Party
leaders can monitor voting in the legislature relatively easily, regardless of
whether the votes are published and transparent. A lack of transparency,
therefore, fosters asymmetry between party leaders and constituents. Voting
transparency is less relevant to citizen monitoring of parties as wholes. Even
when individual voting records are unavailable, those outside the legislature
can get access to information about the policy positions and actions of major
collective actors – parties, unions, business associations – at relatively low cost
through newspapers and broadcast media. This chapter emphasizes the con-
nection between transparency and individual accountability, contending that,
when votes are not readily visible, the costs of information about individual
legislative behavior are prohibitive for legislative outsiders.
To make the case, I address a number of specific questions: What in-
formation about votes is available? What factors generate change in the
revealed information about votes? What conditions are necessary for vot-
ing records to be politically salient? What effect does public voting have on
the relationship between individual representatives and their parties? As
the chapter discusses, the amount of information revealed about votes
varies tremendously across legislatures. Vote records are potentially salient
in all legislatures, although more in single-member-district than multi-
member-district electoral systems. Lawmakers disagree on whether public
voting is desirable, with those who control the agenda generally opposed.
Pressures from opposition and dissident lawmakers and from pro-
transparency actors outside legislatures, as well as technological advances,
all encourage publication of legislative voting records. Finally, there is
a tension between public voting by individual legislators and disciplined
voting among political parties.
44
Visible Votes and Accountability
Country Data on the Number Archival and Interview Archival Data Based
of Visible Votes Data Based on on Research
Field Work Assistance and
Correspondence
Argentina X X
Bolivia X X X
Brazil X X
Chile X X
Colombia X X X
Costa Rica X X
Ecuador X X
El Salvador X X
Guatemala X X
Mexico X X X
Nicaragua X X
Panama X X
Peru X X
United States X X
Uruguay X
Venezuela X X X
What are the implications of examining this particular set of cases? First,
all these cases are presidential, and together they constitute most of the
pure presidential democracies in the world. In these regimes, the separation
of powers allows legislative performance to be evaluated independently
from executive performance (Cox 1987; Diermeier and Feddersen 1998;
Samuels 2004). Because legislative votes on individual policy proposals do
not directly affect the survival of the executive in office, as they may under
parliamentary and hybrid regimes, accountability for individual voting deci-
sions ought to be more pronounced under presidentialism. If public voting is
politically salient anywhere, then it should be so in the Americas. In this
sense, the countries examined in Chapters 3 and 4 may be biased toward high
demand for legislative transparency. Amplified demand could facilitate the
job of detecting conditions that encourage transparency – for example, by
heightening sensitivity to whether particular electoral rules or procedural
rules favoring the opposition favor transparency of individual legislators’
behavior. If this is the case, it may be that in parliamentary and hybrid
regimes the effect of similar conditions on transparency should be muted.
45
The Supply of Visible Votes
1
The discussion in Section 5.4 of data included in the analysis of recorded voting in Chapters
5 and 6, which extends beyond the Americas and beyond pure presidential regimes, elab-
orates further on how research resources affect access to data on legislative voting, and with
what implications.
46
Visible Votes and Accountability
2
This is not meant as a criticism of rhetorical ideal point estimation, which is a ground-
breaking innovation, but merely to note that it is a distinct, and complementary, endeavor
from measuring voting unity.
47
The Supply of Visible Votes
48
Who Can Monitor Votes?
49
The Supply of Visible Votes
50
The U.S. Experience
51
The Supply of Visible Votes
52
The U.S. Experience
53
The Supply of Visible Votes
many House officer elections early in the century. Yet, the effect of public
voting on party discipline, particularly for the highly salient votes to elect
House Speakers, was ‘‘exactly the opposite in the long term . . . [because] the
daylight that shone on speakership elections highlighted regional animosities
just as much as partisanship. It became more difficult to elect Speakers and
organize the House than before the onset of viva voce voting’’ (Jenkins and
Stewart 2003:504–5). The viva voce episode illustrates that, in the U.S.
context, the move away from unmonitored votes initially strengthened the
influence of national party leaders over rank-and-file legislators, but univer-
sal monitoring ultimately strengthened an even greater force, countervailing
that of party – constituents, with their diverse regional demands.
The league, moreover, was not satisfied with fair-sounding pledges and
relied on methods of monitoring recorded votes that echo those of modern
interest groups. ‘‘Elaborate indexes of politicians and their records were
kept at Washington and in most of the states, and professions of sympathy
were matched with deeds. The voters were constantly apprised of the
doings of their representatives’’ (Odegard 1928:91). The Farm Bureau,
the American Legion, the American Medical Association, and the National
Rifle Association engaged in similar activities during this same era (Kile
1948; Mayhew 1974:66–67).
54
The Supply of Recorded Votes in Latin America
55
The Supply of Visible Votes
56
The Supply of Recorded Votes in Latin America
57
The Supply of Visible Votes
By request, electric
Mexico House 500 6 legislators (Art. 148) 155
Mexico Senate 128 6 senators (Art. 148) 156c
Panama Unicameral 71 Majority of those present 144d
(2004) (Art. 196)
Brazil House 513 6%, or party leaders >68e
representing 6%
of members (Art. 185)
Argentina House 257 10% of deputies present 17f
(Art. 190)
United States House 434 20% of quorum (Rule XX) 559
Brazil Senate 81 Majority of those present 125g
(Art. 294)
Venezuela Unicameral 165 Majority of those present 0h
(Arts. 120, 125)
By request, manual
Guatemala Unicameral 140 6 legislators (Art. 95) 8.4
Ecuador Unicameral 100 10% of legislators (Art. 70) 4.5
Bolivia House 130 Majority of those voting 0
(Art. 107)
Bolivia Senate 27 Majority of those voting 0
(Art. 116)
Uruguay House 99 One-third of those present <1.0i
(Art. 93)
Argentina Senate 72 Majority of those present 21j
(Art. 205)
Colombia House 161 Majority of those present 2.5
(Art. 146)
Colombia Senate 102 Majority of those present 2.0
(Art. 146)
58
The Supply of Recorded Votes in Latin America
59
The Supply of Visible Votes
60
The Supply of Recorded Votes in Latin America
61
The Supply of Visible Votes
40
Bolivia
Chile
Costa Rica
Mexico Peru
El Salvador
20
Venezuela
Individualism
Argentina
Uruguay
0
Guatemala
Colombia
-20
Nicaragua
Ecuador
-40
0 5 10 15 20 25
sqrtVV
Figure 3.1 Recorded votes (square root) and preferences for legislative
individualism.
62
The Supply of Recorded Votes in Latin America
the next chapter explores the sources of support for, and opposition to,
visible voting more generally.
3.4.3.1. Peru. The prospect that visible voting would increase the
premium on individual legislator accountability was explicitly on the
minds of Peruvian legislators on both sides of the reform debate in 1998,
as they considered the implications of switching from the traditional hand-
raising method of voting to the electronic system, which had recently been
installed as part of a broader government modernization plan. On September
24, in an effort to embarrass the pro-Fujimori majority on a motion related to
a corruption investigation, the opposition demanded that the electronic
voting machines be incorporated into standard legislative procedure:
The whole reason for electronic voting is so citizens know how their representatives
voted, so [votes] can be publicly justified. It’s an instrument of democracy and trans-
parency, which is why Congress spent as much as it did [to have it installed], not so we
can use it on some votes and not on others. . . . What the country is going to notice is
that the parliamentary majority is afraid that, through the Internet and other mech-
anisms, its votes on some matters will be made visible. (Congreso del Perú 1998)
63
The Supply of Visible Votes
3.4.4. Why More Votes Are Not Recorded, and Why Recorded Votes
Are Not Always Visible
Even if we take technological and procedural barriers into account, the
Latin American legislatures generally record fewer votes than the U.S.
Congress. After all, the U.S. Senate has no electronic voting, and House
rules impose a significant request threshold, yet both record hundreds of
votes per year. One obvious explanation is a lack of staff resources in Latin
American legislatures without electronic systems. Legislatures in the
region are chronically and notoriously underfunded, and recording votes
by hand is labor-intensive and time-consuming.
Procedural costs cannot be the whole story, however. Even where elec-
tronic systems are in place in Latin American legislatures, their use is not
a given. Electronic systems are in place in the Costa Rican and Venezuelan
assemblies, but they are never used, just as in Panama until 2004, and the
electronic systems in the Argentine and Colombian lower chambers are
very rarely employed. In other cases, the systems are used regularly, but
voting records are not systematically published. The Nicaraguan Assem-
bly, for example, records all votes, but does not publish the records. The
Argentine Chamber’s Web site puts up the aggregate (yes, no, abstain,
absent) results for votes taken electronically but includes the lists of in-
dividual deputies’ voting decisions only sporadically. The Mexican Senate’s
Web site, curiously, publishes its recorded votes the evening of each given
session but then removes the records when the votes from the next session
64
Conclusion
are put on line. The Mexican Chamber’s Web site has changed its policies
for making recorded voting data available various times during the early
years of the 2000 decade before settling on putting up complete records of
electronic votes.3
Public access to information about electronically recorded votes may be
partly attributable to technical capacity. Maintaining a comprehensive
Web site taxes the resources of many assemblies. On the other hand, party
leaders and the members of dominant coalitions sometimes prefer not to
make voting records public even when they are kept and not to use elec-
tronic voting systems even when they are in place. For now, it is worth
noting that, although the point of departure is different, the relationship
between electronic systems and recorded voting in Latin America runs in
the same direction as that in the United States, albeit starting from a dif-
ferent point. In the United States recorded voting is common even in the
absence of electronic voting, and it became more prolific with the adoption
of an electronic system in the House of Representatives. In Latin America,
recorded voting is negligible in the absence of electronic voting. It is in-
creasingly common – but not a given – where electronic systems reduce
procedural costs.
The experiences of individual countries that have adopted electronic
voting systems affirm the pattern evident in the cross-national data: elec-
tronic voting and minimizing procedural barriers to recorded voting boost
the amount of information available to those outside the legislature about
legislative decision making.
3.5. Conclusion
Legislative voting can be an element of accountability only if votes are
publicly known. They can be known, in turn, only if they are recorded
and the records are available. Voting records are so essential to how legis-
lative accountability is conceived in the United States, and have been for so
long, that it is easy to forget that there is nothing automatic about their
relevance, or even their existence and availability. Beyond the United
States, essential components of the legislative voting–legislative account-
ability relationship are often missing.
3
The practices described in this paragraph refer either to convention at the time field re-
search was conducted for this project, during the first years of the 2000 decade or, in the
references to Web site content, to practice up through the middle years of the decade.
65
The Supply of Visible Votes
66
Appendix
even when nominales are taken, and instead write down their votes and pass
them forward. The complete record of any nominal, therefore, is actually
assembled by the chamber’s Division of Recording and Dissemination.
Having established exactly what I was seeking, the technical assistant to
the secretary-general took me to the director of the Division of Recording
and Dissemination. This director listened to my request for electronic
records of votes, and then sent me, with his assistant, to the deputy secretary
of the House chamber, who, I was assured, was the guy who knows where
the computer files containing votes are, and how they are organized.
The deputy secretary listened to my request and told me that records of
all nominales are published in the Gaceta (the official published record of
legislative proceedings), which I already knew. I explained that I hoped to
avoid having to go through every page of every Gaceta to find the rare
nominales, make photocopies, and then transcribe each vote to a spread-
sheet. I told him that I had assumed electronic records must exist – perhaps
even of more votes than the formal nominales – given the existence of an
electronic voting system. I noted that even the records published in the
Gaceta are written up on computers prior to printing, that electronic copies
would be far more convenient than print versions. And I pointed out that
various sources – legislators and staff – within the House of Representatives
itself had confirmed the public availability of these resources. The deputy
secretary countered that the electronic voting system does not produce
a record of the votes; then clarified that the system does produce a record,
but that the record is deleted from the system immediately after each vote is
taken and the result of it is printed up. Subsequently, he explained, the
Division of Recording and Dissemination transcribes from these printouts
a new list of who voted how, which is inserted in the Gaceta. The sub-
secretary explicitly did not offer to make electronic records of the Gaceta
available. Rather, he reiterated his offer to let me – or anyone, because this
information is public, after all – leaf through the hard copies page by page
to find the votes and make photocopies.
After leaving the deputy secretary’s office, the assistant from the
Division of Recording and Dissemination (still with me) told me that the
director of the Legislative Archive, or possibly director of the Gaceta, would
likely be more cooperative, and that I should check with them. I took this
advice, but with results that should, by now, be predictable. Pursuit of legis-
lative voting records in many other countries, including Bolivia, Ecuador,
and El Salvador, yielded similar results.
67
4
1
For a contrary argument, that even citizen voting at the polls should be public, see Brennan
and Pettit (1990).
68
Is Transparency Desirable?
2
For example, Snyder and Ting (2005) offer a formal model to explain why public voting in
legislatures ought to be appealing both to legislators and to citizens. The outcome depends
on symmetrical monitoring of legislators’ voting behavior by both their constituents and
some other principal – say, a powerful interest group or a political party leader – whose
preferences may be at odds with constituents’. Symmetrical monitoring is necessary for
legislators to make binding electoral contracts with their constituents – to commit credibly
to be faithful representatives and to expect any reward for doing so. A critical assumption of
the model is that information about legislators’ votes gets to constituents.
69
Demand for Visible Votes
3
Although Burke would eventually prove to be among the most sympathetic members of the
British Parliament to American grievances against the crown, many of the colonists did not
share his views about deference to legislators. One particularly un-Burkean reform ad-
vanced by reformers in Pennsylvania around the eve of independence would have required
any proposed law to be posted in public houses for a year before the colonial legislature
could vote on it, to ensure citizens could monitor their representatives in advance of acting
(Morone 1990:40). The Internet notwithstanding, these Founders appear to have had
a keener sense than we do at present of how to use available technology to foster account-
ability.
70
Incentives to Monitor and Publicize Votes
4
Open lists were used in Colombia beginning with the 2006 election, but the previous system
of multiple lists was at least equally personalistic (Cox and Shugart 1995).
71
Demand for Visible Votes
72
Incentives to Monitor and Publicize Votes
73
Demand for Visible Votes
74
How the Political Actors See Things
5
Legislative parties in Latin America are referred to variously as bloques, fracciones, bancadas,
grupos, or partidos. For simplicity, I refer to such units generically as party groups.
75
Demand for Visible Votes
the other hand, have 250 to more than 500 members, and all three have
electronic voting systems installed. Chamber membership is positively
correlated (.45) with use of electronic voting, but the correlation between
membership and frequency of recorded votes is much smaller (.15), and not
significant. Party leaders in all chambers where interviews were conducted
emphasized that discipline is expected in voting, so recording individual
legislators’ votes would be redundant (R. Alvarenga and Duch interviews).
According to Gonzalo Sánchez de Lozada, former president of Bolivia,
‘‘Here it’s not like the United States where you say ‘I voted this way or
that.’ . . . Here, people presuppose and expect legislators to be loyal to
their party.’’
Overall, party leaders interviewed were consistently dismissive of the
need to record and publish legislative votes. Press reports from other legis-
latures confirm this pattern. The president of the Panamanian National
Assembly dismissed a proposal by opposition deputies to adopt electronic
voting on the grounds of unspecified technical problems with the system
and the lack of training in its use among the deputies themselves (Tapia G.
2004). After an intervening election, however, the next Panamanian As-
sembly saw fit to begin recording and publishing votes (Asamblea Nacional
de Panamá 2005). Similarly, in 2003 party leaders in the Texas House of
Representatives initially resisted demands to record and publish votes on
the grounds that to do so would be too costly in terms of time, despite the
fact that an electronic voting system was already in place, and in terms of
the extra paper required to include the voting rolls in the published House
Journal (Dallas Morning News 2003a). Neither factor proved pivotal in the
longer run, however; by 2005 Texas votes were available on the state leg-
islative Web site (Texas Legislature Online 2005). In contrast to leaders,
legislators further removed from positions of control over the agenda ex-
press consistent support for recorded voting. Many of these also suggested
that party leaders are actively hostile, rather than indifferent, toward
recorded voting.
76
How the Political Actors See Things
voting results, to hold their adversaries’ feet to the fire on votes that are
unpopular, or to publicize dissidence within the majority party or coalition.
The truth is the following: the president of the Congress often manipulates the
votes. So, when you don’t have a nominativa . . . if the secretary says there are sixty-
four votes out of the 110 delegates who are present in the hall, the article is passed.
[My party] usually has two of its assistants in the Congress at the front, on both sides
of the plenary hall. . . . It’s a warning for the secretary, because on more than one
occasion we’ve demonstrated with the accounting that we have brought with us that
they are giving a result that’s incorrect. [See also Lucero interview.]
By Alvarenga’s estimation, sufficient time has now passed that the Salva-
doran Assembly should adopt electronic voting, but to this point it has not.
It is worth noting that Alvarenga’s PDC was a majority party during the
period when he found ‘‘contriving votes’’ acceptable practice, but currently
it is much smaller and outside government.
Legislators interviewed in every chamber that lacked electronic voting
asserted that outcomes were altered when signal votes, which generally
involve standing, hand raising, or banging on the table (el pupitrazo), were
77
Demand for Visible Votes
used, and they generally expected that electronic voting would remedy the
problem (interviews with Landazuri and Lucero in Ecuador, Sánchez Bez-
raı́n in Bolivia, Navarro in Colombia, and Guido in Costa Rica). Nicara-
guan deputy Marı́a Lourdes Bolaños (FSLN) confirmed the improvement
under electronic voting on the basis of recent experience:
I think the change is transcendental. Members of the Salvadoran opposition have
told me they want to acquire an electronic system because they regard transparency
as very weak in El Salvador, to the point where the Junta Directiva manipulates
votes. They always overcount, they’re never satisfied. In contrast, we are satisfied
with the votes. We believe there is transparency, we believe there’s efficiency.
That’s important. With manual voting, for all the time it would cost us, now we
have agility. It’s not just about transparency, but agility.
78
How the Political Actors See Things
requested in any instances either where there is doubt about the outcome or
where enough legislators want to insist on a public record (Acosta, Carva-
jal, and Lucero interviews). Yet opposition legislators objected that those
who control the flow of legislative traffic fail to handle such requests even-
handedly (Devia, Garcı́a Valencia, and Holguı́n interviews). According to
Colombian senator Rafael Orduz,
4.3.2.4. Party Mavericks. Making votes public makes it easier for legis-
lators to stake out positions independent from those of their parties.
Recorded votes can serve as means for maverick legislators to ‘‘go public’’
over the heads of party leaders and, in so doing, to establish reputations
either among a target audience of supporters or perhaps nationally. The
rare decisions to hold nominales in systems where anonymous legislative
voting is the norm can illustrate this. According to Costa Rican minority
leader Sibaja (interview),
One sign that there’s going to be a nominal isn’t that the opposition is divided –
that’s no problem. The problem is when the governing party is divided. There was
a famous case here in the early 1970s, having to do with student protests over an
agreement that permitted a transnational company to mine [in a wilderness area]. It
was called the Alcoa Agreement. At that time, the PLN controlled the presidency
and had a big parliamentary majority. One government deputy started the fight.
That deputy himself later became president, but not as a member of the PLN – don
Rodrigo Carazo Odio, who founded the Unity Party, which is governing currently.
He led a group of PLN deputies to break the party line. I think that was the last time
they used a nominal on an important issue, precisely because the government’s
fraccio´n divided at that moment. That was thirty years ago. It’s not common.
79
Demand for Visible Votes
party mavericks with a forum for position taking and insofar as they open
legislators to demands of accountability for their votes from actors outside
the chamber (Bolaños interview).6
In the first years of this century, a similar story was unfolding in Nicar-
agua, where a group of FSLN deputies objected to a deal cut between their
leaders and those of the president’s Liberal Party on a package of consti-
tutional and electoral system reforms. The FSLN dissidents took advan-
tage of the recently adopted electronic voting system to publicize their
rebellion, drawing the ire of loyalists, such as Deputy Maria Lourdes Bola-
ños: ‘‘[The voting records] have been used to make public the divisions
within coalitions. Not for transparency, but to the advantage of those four
deputies who are against the pact. That’s what it has come to’’ (Bolaños and
Urbina interviews). Bolaños evidently saw no boost to transparency inher-
ent in making votes visible. One of the dissidents, Deputy Mónica Balto-
dano, however, was becoming acutely aware of the costs of going public
with her breech of voting unity: ‘‘We broke discipline. . . . So [the party
leadership] ruled that whoever did not accept party decisions could not
aspire to electoral posts. Everyone knew I wanted to run for mayor of
Managua, and this way I couldn’t be nominated. It’s almost certain that
they won’t permit me to run for reelection as a deputy either.’’ As antici-
pated, Baltodano was subsequently barred from nomination for reelection
as deputy, citing as the reason her vote in the Assembly against the electoral
reform law. By 2002 she was out of elected office.
6
Testing this proposition empirically – for example, by comparing party unity levels in
legislatures with electronic voting against levels on recorded votes in legislatures without
electronic voting – is problematic, because votes that get recorded in the latter may be
biased toward disunity. In the Alcoa vote described by Alexis Sibaja, for example, both
major parties split.
80
How the Political Actors See Things
4.3.4. Presidents
Presidential commitment to recorded voting is a product of the specific
political conditions at hand and the goals of a particular president. The rare
circumstances that land the issue at the top of a presidential agenda, how-
ever, may be sufficient to establish recorded voting as standard practice.
As with party leaders, interview subjects dismissed the need for execu-
tives to rely on recorded voting to monitor their legislative allies, on the
grounds that informal networks within legislatures themselves were
81
sufficient for these purposes (Guerra and Holguı́n interviews). Neverthe-
less, presidents may demand recorded voting for other reasons – as a gambit
to enfranchise third-party monitors to offset the inherent advantages of
legislative party leaders, or even out of a genuine desire to increase trans-
parency in the policymaking process.
Many presidents express a generic interest in modernization and effi-
ciency and a willingness to push such demands on a reluctant legislature.
An ironic example is the case of Alberto Fujimori, whose administration is
not generally associated with transparency. Yet Fujimori’s campaign to
modernize the state included an initiative to computerize the Peruvian
Congress, which in turn included the installation of electronic voting
machines (Cevasco Piedra 2000). Although Fujimori’s legislative allies
initially refused to use the equipment, the legislative opposition eventu-
ally succeeded, through the aggressive use of obstructionist tactics, in
forcing the adoption of electronic voting as standard operating proce-
dure (Carey 2003). Thus, Fujimori’s modernizing drive appears, un-
intentionally, to have produced the regular practice of recorded voting
in Peru.
Less inadvertently, and also less successfully, Colombian President
Alvaro Uribe’s first act as president in 2002 was to introduce a broad pack-
age of political reforms, the first element of which was a requirement that
all votes taken in the Colombian Congress be recorded and made public (El
Tiempo 2002). The priority Uribe gave to this procedural detail is remark-
able given that he assumed the Colombian presidency in the midst of a civil
war.7 His stated rationale was that lack of confidence in government insti-
tutions was responsible for the crisis of the Colombian state and that trans-
parency was needed to produce greater accountability among elected
officials.8 Facing congressional resistance to his proposal in 2002, Uribe
put his reform package directly to voters in a fifteen-point referendum the
next year. The public voting provision received 94 percent support
among votes cast, but only 24.7 percent of eligible voters participated. In
Colombia, referenda require 25 percent participation to be valid, so Uribe’s
proposal failed, and most votes in Colombia remain invisible to those out-
side the Congress.
7
Uribe’s inauguration ceremony itself was subject to a mortar attack.
8
It is worth noting that Colombia’s previous president, Andres Pastrana, in the name of
transparency, had also tried to pass a package of reforms that included the requirement of
public voting. The proposal died in the face of legislative opposition.
82
Effects of Recorded Voting
4.4.1. Transparency
When asked the open-ended question, ‘‘What effects, if any, does recorded
voting have on legislative representation?’’ most legislators mentioned an
increase in transparency.9 The term, however, is sufficiently general (and
such talk is sufficiently cheap) that it is worth spelling out more explicitly
what it entails. In its crudest sense, the transparency afforded by recorded
voting is a check against the ability of legislators to lie outright about the
policies they have supported or opposed inside the chamber. None of my
interview subjects confessed to having perjured themselves in this manner,
but claims that their colleagues had were common.10 Colombian senator
Rafael Orduz (interview) was willing to name names:
I’ll give you an example, having to do with a particular part of a recent tax reform. In
public – I’m talking about on television – the president of the Liberal Party, Luis
Guillermo Veles, said he was against it. In the vote, inside the Senate, he voted in
favor. There was no TV and no recorded vote, but nobody has called him on it in
public. So publicly, he continues to position himself as if he had opposed the article
I’m talking about.
Orduz’s point was that a recorded vote on the tax measure in question would
have offered a deterrent against the obfuscation of which he accuses Senator
Veles, or else have provided incontrovertible evidence with which any of
Veles’s opponents (presumably including Orduz) could expose his perfidy.
Another Colombian senator offered a similar assessment with respect to
why the Colombian Senate never used an electronic voting system that
been installed two years before: ‘‘They say it has technical problems, but
9
In legislatures that do not regularly record votes, I asked, ‘‘What effects, if any, do you
expect recorded voting would have on legislative representation?’’
10
I am still not sure how I managed to overselect honest politicians to such an extreme.
83
Demand for Visible Votes
this is just an excuse because they don’t want votes to be public. Too many
senators are afraid they will lose votes if they are all made public’’ (Blum
interview). I pursued the same issue later, at a meeting attended by a group
of senators and representatives, asking why neither chamber of the Colom-
bian Congress used its electronic voting system:
Anonymous senator: ‘‘There were problems. They didn’t work.’’
Author: ‘‘Why not? Technical problems?’’
Anonymous senator: ‘‘Well, politico-technical problems.’’
All legislators: [Big laughs all around the table.]
84
Effects of Recorded Voting
One example, drawn from the Texas state legislature, suggests such an
effect for recorded voting. In April 2003 the state House of Representatives
was considering a motion to kill a widely popular proposal to require
legislative candidates to disclose the sources of their campaign contribu-
tions. The motion was initially put to a nonrecorded vote and appeared to
be headed toward passage (which would have killed the bill); however,
‘‘When a recorded vote was requested, the scoreboard changed completely
and the motion to kill the bill failed’’ (Dallas Morning News 2003b).
Multiple independent interviews conducted in Peru pointed toward
another case where recorded voting changed an important policy outcome
(Ması́as, Pease, De Althaus, and Ortiz de Zevallos interviews). At issue was
a proposed change in the electoral system for the 2001 election. Incumbent
legislators had been elected from a single national district. The proposed
reform would divide Peru into multiple electoral districts (circunscripciones)
defined by departmental boundaries. Despite popular support for the idea,
many legislators were skeptical about altering the rules under which they
had, by definition, been successful. Congressman Henry Pease (Union por
el Peru) provided the most compelling account of recorded voting’s effect
on the outcome:
[This reform] obviously was not good for small parties, or for those that knew
that, after the way they had governed, they were going to be small parties. There
was a lot of tension when this issue was put to debate, with strong public opinion
in favor, and a bunch of legislators demanded, based on an article of the rules,
that the vote should be secret. I wasn’t in Congress at the moment because I was
sick in the medical clinic, so I saw on television the impact, above all, of public
opinion. I was in the clinic at least from 6:00 or 7:00 P.M. on, in a room watching
the TV, and the nurses were coming in – not to look after me, but to watch the
TV and express their indignation at what was happening, because as soon as they
saw that it was going to be a secret vote they said ‘‘It’s going to lose’’ and in fact,
they didn’t get enough votes to get rid of the national district. This led to
a mobilization and to demands of all sorts and criticisms of all sorts and allowed
us to force, a month later, another proposal, and vote on it . . . and finally,
even though it was a much more radical bill, it was accepted because of public
pressure.
85
Demand for Visible Votes
more bluntly, even if votes are visible, does anyone look? The Peruvian
fieldwork for this project was conducted in May 2001 – a period of intense
politicization in that country, after the fall of the Fujimori regime, and in
between the two rounds of the election that produced the presidency of
Alejandro Toledo. Congress, which had produced the interim president
Valentı́n Paniagua, and which was conducting investigations into the spec-
tacular corruption charges against members of the Fujimori administra-
tion, was in the national spotlight. In this context, the Peruvian media
reported intensively on congressional voting records. Newspaper articles
reproduced roll calls (Diez Canseco 2001; La Gaceta 2001), and television
talk shows focused on motivations (De Althaus interview). Recorded voting
in Peru appeared to have, in very short order, established itself at the core of
political discourse.
One should be cautious, however, about generalizing too quickly from
the experience of Peru in 2001, which was extraordinary on a number of
counts. The early experience of Nicaragua with recorded voting stands in
contrast to Peru. Field research in Nicaragua was conducted in August
2000, and electronic voting had been adopted as standard operating pro-
cedure only eight months earlier, in January. Deputy Carlos Hurtado
(Accion Conservador) described the status of the voting records this way:
Hurtado: Despite the fact that votes are recorded, they aren’t widely known
among the people, except when a particular issue becomes decisive at
election time.
Author: Is this a process that is beginning?
Hurtado: Yes, it’s happening. It requires that the electorate, the political analysts,
have the record. It’s not so simple to create this record because it
requires a certain infrastructure, a certain culture, a certain systematiza-
tion. There’s no independent center that keeps a record of the votes. In
the United States, there is – there are lots that keep complete records of
the details. That’s more sophisticated. I think eventually we’ll get there,
but certainly as of now it’s not so easy. At least it’s known when a certain
deputy takes a certain position.
86
Effects of Recorded Voting
11
I found no relationship between public votes and ‘‘confidence in Congress,’’ as measured
by the Latinobarómetro during the late 1990s through early 2000s. To the extent that the
crafting of legislation resembles sausage making, per Bismarck’s famous observation, vis-
ibility may enhance accountability and deter corruption without necessarily increasing
public affection for Congress.
87
Demand for Visible Votes
Figure 4.1 Recorded votes per year (square root), by country, and 2004 Trans-
parency International Corruption Perceptions Index.
88
Effects of Recorded Voting
This analysis cannot test the hypothesis that recorded voting weakens
discipline, for the obvious reason that no data exist and no levels of party
voting unity can be measured when voting records do not exist, but the
interviews support this proposition. Deputy Mónica Baltodano, the FSLN
dissident sanctioned for voting against the party, described strong public
support for independence from absolute party discipline: ‘‘There is a great
tolerance [of indiscipline] among citizens and Nicaraguan society, which
can be demonstrated in polls and studies. But within the political institu-
tions, there’s great intolerance, above all within the party.’’ A vignette
provided by Peruvian congressman Carlos Blanco Oropeza (Cambio 90 –
Nueva Mayoria) illustrates the expected effect among party leaders of
recorded voting on discipline in more colorful terms:
I’ll never forget when, in 1998, I was vice president of the Congress, and we hosted
some German legislators, and naturally one of the things we did was to visit the
facilities of the Congress. I accompanied them to the chamber, the place where we
meet, and right there is the screen for the public votes, and I’ll never forget the
German legislator – who was a leader of his party – and he said to me, ‘‘Is the vote
public?’’ So I explained, and he said, ‘‘You guys are crazy. How can you control the
members of your party? Everyone has to vote how the party votes.’’
89
Demand for Visible Votes
90
The Trend toward Visible Votes and Its Limits
91
5
Counting Votes
92
Party Voting Unity and Collective Accountability
then the party’s label has no informational value. Third, unity affects the
ability of parties to win votes and shape policy. Unity determines whether
governments can act decisively or, by contrast, whether each legislative
decision requires separate deliberation and the construction of a distinct
support coalition. In this sense, party unity is linked to the ability of parties
and governments to deliver the promises in their platforms (Bowler,
Farrell, and Katz 1999).
I focus on three main characteristics of how parties vote: how consis-
tently their members take the same position on the motions on which they
vote, how much they win, and how frequently their losses might have been
avoided but for breaches of voting unity. Voting together matters not just
because mobilizing its full voting capacity can help a party win floor votes
and so promote its supporters’ interests. Voting together (or failing to do
so) also sends information to citizens about the party’s policy positions and
its level of commitment to them. When votes are visible, cross-voting
within a party or failure to mobilize its members on a given measure blurs
the party’s brand name. A party that mobilizes its potential and votes in
a unified manner, by contrast, fortifies and clarifies its reputation. Thus, the
degree to which copartisans vote together is relevant to accountability both
through the delivery of wins and losses and through its communicative
content.
The indices developed in this chapter measure raw levels of party unity
and are agnostic as to whether the source of unity is like-mindedness among
legislators (cohesiveness) or pressure from party leaders (discipline). Raw
levels of party unity matter in their own right. Low unity reduces the com-
municative value of party labels to voters, and high party unity increases it,
regardless of whether the source is cohesiveness or discipline. Low unity
reduces the ability of parties to deliver their policy promises through legis-
lation, and high party unity increases it, regardless of whether the source is
cohesiveness or discipline. Indices of party unity that can be deployed across
systems and across parties describe variance in a common metric that makes
cross-national comparison possible. As this chapter demonstrates, party
unity varies a lot, both across political systems and among parties within
systems. The next chapter uses the indices developed here to shed some
light on the relative importance of cohesiveness versus discipline in account-
ing for the levels of party unity observed empirically and to test propositions
about the sources of high and low party unity.
Party unity is a necessary condition for collective accountability, not its
guarantee. As Chapter 2 highlights, when voters are unable to demand
93
Counting Votes
where aye and nay are calculated as proportions of those voting either aye or
nay, and so sum to 1.0. The RICE score can range from zero (equal numbers
vote aye and nay) to one (all members who cast votes vote together).
One limitation of RICE is that it does not account for nonvoting, levels
of which are substantial in most legislatures. For example, if a party has 100
members, 60 of whom cast affirmative votes on a measure and 40 of whom
94
Measures of Voting Unity and Success
abstain or otherwise do not vote, RICE regards this event as perfect unity
(RICE ¼ 1.0), equivalent to if all 100 members of the party voted aye (or
nay). Intuitively, these are two fundamentally different events, the differ-
ence between which could obviously affect the vote outcome. This suggests
some type of measure that is sensitive to whether copartisans vote, as well as
to how they vote. For this, I propose UNITY, which captures the extent to
which a party exercises its decisive capacity on a given vote:
95
Counting Votes
score from each vote according to how closely the vote was contested. The
procedure for calculating weighted RICE (WRICE) and UNITY (WUNI-
TY) indices is described in Appendix 2.
96
The Silence of Nonvotes
97
Counting Votes
98
The Silence of Nonvotes
Figure 5.1 WRICE, WUNITY, RLOSER, and ULOSER indices for Argentina,
1995–97.
Figure 5.3. The drop-off from WRICEi to WUNITYi was more pro-
nounced for President Toledo’s Peru Posible (PP) party than for others,
and it cost the government party victories on about 3 percent of all
recorded votes – a higher rate than for any other party.
Along the same lines, consider the voting record of the Czech Republic’s
parliament during its 1996–98 term. In Figure 5.4, the parties of the gov-
erning coalition are white and all others shaded various hues of gray. There
is the familiar drop-off from WRICEi to WUNITYi indices, and the rise
from RLOSERi to ULOSERi, as the latter capture nonvoting as well as
cross-voting. But the rise in ULOSERi is spread across all parties and is
more pronounced within the governing coalition than outside it. In short,
patterns of these indices across countries suggest that nonvoting equilibria
are possible but not uniform. Moreover, the combined patterns of
WRICEi, WUNITYi, RLOSERi, and ULOSERi across parties can detect
signs of such equilibria in some cases and help rule them out in others.
99
Counting Votes
100
The Silence of Nonvotes
Figure 5.3 WRICE, WUNITY, RLOSER, and ULOSER indices for Peru, 2001.
101
Counting Votes
Country: Czech Republic. Legislature: 1-Jul-96 Country: Czech Republic. Legislature: 1-Jul-96
1
.75
.75
.5
.5
.25
.25
0
0
.2 .4 .6 .8 1 .2 .4 .6 .8 1
Proportion of Seats in Chamber Proportion of Seats in Chamber
Country: Czech Republic. Legislature: 1-Jul-96 Country: Czech Republic. Legislature: 1-Jul-96
0 .05 .1 .15 .2 .25
.2 .4 .6 .8 1 .2 .4 .6 .8 1
Proportion of Seats in Chamber Proportion of Seats in Chamber
Figure 5.4 WRICE, WUNITY, RLOSER, and ULOSER indices for the Czech
Republic, 1996–98.
as well as corrections for bias inherent in RICE and UNITY scores for
small parties, are discussed in Appendix 6.
102
Data on Recorded Votes
1
For this, legislative scholars owe gratitude to sustained effort and a commitment to data
sharing by Professor Keith Poole of the University of California, San Diego.
2
One exception was Israel, where an Israeli colleague put me in contact with a Knesset staffer
who had access to archives and was willing to transcribe hard copy records of votes into
spreadsheet format. Meanwhile, legislative scholars with diverse language skills and insti-
tutional contacts should make it a priority to determine where any yet-untapped archives of
recorded legislative votes are, and to collect and disseminate those data. To facilitate this
effort, VoteWorld: The International Legislative Roll-Call Voting Website (http://ucdata.berkeley
.edu:7101/new_web/VoteWorld/voteworld/) is a data clearinghouse established in collabora-
tion by scholars committed to open access. All the data for this project are available either via
VoteWorld or at http://www.dartmouth.edu/~jcarey.
103
Counting Votes
next includes all those where I know data to exist, making this, to my
knowledge, the broadest cross-national study of legislative voting under-
taken to date.
The samples of votes from these legislatures are shaped by vote avail-
ability and resource constraints, such as the labor to transfer each vote to
machine-readable format. Some assemblies post vote records on line as lists
of names that, with some work, can be prepared for analysis. In other cases,
my travel to the assemblies themselves aimed at locating and collecting
hard copies of whatever votes were recorded, or else contracting with local
assistants to collect the information. Scholars who had collected vote data
in similarly painstaking fashion were, by and large, generous in making
those available. In some cases (e.g., United States, Uruguay), my sample
of votes represents all recorded votes during complete legislatures; in
others (e.g., Chile, Israel, Peru), it includes all votes between specified
dates; and in still others (e.g., Canada, Czech Republic, Russia) the sample
includes those votes that colleagues made available.
What are the implications for how we interpret measures of voting
unity? If unity is fundamentally different in legislatures where recorded
votes are unavailable, then the measures reported here may be unrepresent-
ative of the whole population of legislatures. If the factors correlated with
higher and lower levels of unity are different in legislatures where recorded
votes are unavailable, then the explanations offered here for levels of voting
unity are limited to environments where votes are visible. Time spent ob-
serving legislative behavior in legislatures both where votes are visible and
where they are not leads me to only a modest conjecture on the possibility of
differences in the power of presidential patronage to sway votes in visible-
versus invisible-vote systems. I discuss these in the following chapter. That
said, it must be noted that where recorded votes are unavailable, we cannot
know for sure what levels of voting unity are and what explains them.
I draw on recorded vote data from lower legislative chambers across
nineteen countries. The unit of analysis is the party group during a given
legislature. I calculated voting indices for each party in each legislature that
it enjoyed representation (%WONi and WUNITYi), or where its group
consisted of two or more legislators (WRICEi and ULOSERi), or three or
more legislators (RLOSERi). The dates, the total number of votes, and
properties of the corresponding mean CLOSEj scores for the cases exam-
ined in this chapter are shown in Table 5.1.3
3
For an explanation of CLOSEj scores, see Appendix 2.
104
Data on Recorded Votes
(continued)
105
Counting Votes
106
Describing Voting Unity
(e.g., aye, nay, abstain, or no vote). Finally, there is variance in the overall
tendency toward consensus or contestation in votes. Mean CLOSEj sum-
marizes the extent to which an average vote was contested for each case.
Votes were most narrowly won in New Zealand, Argentina, the French
Fourth Republic, and Guatemala, less so in Ecuador, Chile, Peru, and
especially the Philippines. In all legislatures, some votes are consensual,
but in most there are deep divisions on many votes as well – enough that we
can be confident that the real fights over policy have not all ended before
votes come to the floor.
4
There are the fewest observations on RLOSERi, because it is not calculated for parties with
fewer than three members. There are more observations on the other indices.
5
Table 5.2 summarizes information on four voting unity indices across all the countries
examined. It is useful as a reference source, but the format does not facilitate comparison
across cases. Figures 6.2 and 6.3 illustrate more clearly the patterns of WRICEi and RLO-
SERi and their variance cross-nationally.
107
Counting Votes
Table 5.2. Voting Unity Index Averages and Standard Deviations by Country
all exhibit low values. Parties in the United States average the highest rates
of losses because of both cross-voting (RLOSER) and failure to mobilize
fully (ULOSER), although relatively high rates on RLOSER are also pres-
ent in Nicaragua, Poland, and Uruguay, and on the latter in various coun-
tries, although the preceding discussion of nonvoting equilibria suggests
wariness toward that statistic. A few countries, including Australia again
plus Canada, Guatemala, and the Philippines, show no – or almost no –
vote losses due to breaches in voting unity.
108
Describing Voting Unity
Figure 5.5 WRICE, WUNITY, RLOSER, and ULOSER indices for Australia,
1996–98.
their party systems. The statistics presented in Table 5.2, for example,
suggest Australia as the prototype of a highly unified legislative party sys-
tem. Figure 5.5 illustrates this uniformity by juxtaposing its WRICEi,
WUNITYi, RLOSERi, and ULOSERi indices during the 1996–98 period.
Again, government parties are white and opposition shaded various gray
hues. The relative simplicity of Australia’s coalition structure during this
period and the regularity of legislative voting behavior are clear.
Compare this with the structure of the Israeli Knesset, shown in Figure
5.6, during a period from October 1999 to November 2000 from which I
collected a sample of votes. First, the far greater fragmentation of the
Israeli party system and governing coalition structure is evident. Israeli
parties also show somewhat more variance in voting unity across the
different indicators, and their overall levels demonstrate less unity and
some vote losses, even among parties in the governing coalition, because
of disunity.
109
Counting Votes
Figure 5.6 WRICE, WUNITY, RLOSER, and ULOSER indices for Israel,
1999–2000.
110
Describing Voting Unity
Figure 5.7 WRICE, WUNITY, RLOSER, and ULOSER indices for Brazil,
1995–1998.
Israeli and Brazilian systems. The relatively low WRICEi indices illustrate
a substantial amount of cross-voting with which even casual observers of
the U.S. Congress are familiar, but the near parity between WRICEi and
WUNITYi also shows that U.S. legislators are diligent about getting to the
floor to vote. Most of the disunity in the U.S. Congress comes in the form
of cross-voting, rather than nonvoting. As a result, while both the loss-
based indices are high for the United States, RLOSER is nearly as high as
ULOSER, especially for the Republicans, the majority party during this
Congress.
Because the voting unity indices developed in this chapter summarize
vast amounts of legislative activity in relatively simple statistics, it possible
to compare distinct legislatures, or the same legislature in different periods,
or both, according to common metrics. This is useful for describing and
visualizing legislative party systems, but the goal of developing these tools
111
Counting Votes
Figure 5.8 WRICE, WUNITY, RLOSER, and ULOSER indices for the
United States, 1995–97.
is to explain the conditions that generate high and low levels of voting
unity, mobilization, and ultimately partisan and government success or
failure on the floor. The next chapter focuses on that task.
112
Appendix 1
scores are ‘‘perfect’’ at 1.00. Voting participation increases across the six
votes in every party – that is, the number of nonvotes declines. Now consider
the pattern of Party A, which experiences increasing cross-voting as more of
its members cast decisive votes; thus, its RICEAj scores plummet faster than
its UNITYAj scores, converging at zero on the last vote. Party B experiences
substantial cross-voting on early votes but pulls together subsequently such
that both RICEBj and UNITYBj rise and converge as more legislators mo-
bilize on later votes. Party C experiences no cross-voting on any votes, so
UNITYCj rises to converge with RICECj as mobilization increases.
113
114
Appendix Table 5.1. Examples of RICE and UNITY Scores, and Weighted and Unweighted Indices, in a Hypothetical Legislature
Tally CLOSEj TallyA UNITYAj RICEAj TallyB UNITYBj RICEBj TallyC UNITYCj RICECj
[150,0,150] 0.00 [50,0,50] .50 1.0 [50,0,50] .50 1.00 [50,0,50] .5 1.00
[162,18,120] .20 [54,6,40] .48 .80 [48,12,40] .36 .60 [60,0,40] .6 1.00
[168,42,90] .40 [56,14,30] .42 .60 [42,28,30] .14 .20 [70,0,30] .7 1.00
[168,72,60] .60 [56,24,20] .32 .40 [32,48,20] .16 .20 [80,0,20] .8 1.00
[162,108,30] .80 [54,36,10] .18 .20 [18,72,10] .54 .60 [90,0,10] .9 1.00
[150,150,0] 1.00 [50,50,0] 0.00 0.00 [0,100,0] 1.00 1.00 [100,0,0] 1.0 1.00
Unweighted
indices .32 .50 .45 .60 .75 1.00
Weighted
indices .20 .27 .55 .60 .87 1.00
Appendix 2
X X
WUNITY i ¼ UNITY ij CLOSEj = CLOSEj ;
where
6
When the threshold for passing a measure is a simple majority of those voting, the formula
can be written as: CLOSE ¼ 1 (2 * |50% %AYE|). However, when passage requires an
extraordinary majority, the more general formula still applies. This form of the general
equation was suggested to me by Jeanne Giraldo.
115
Counting Votes
7
In a handful of legislatures, such as Brazil’s Chamber of Deputies and the U.S. Congress on
some votes, parties’ formal positions on specific measures are reported as part of the assem-
bly’s published record. This practice, however, is sufficiently rare as not to be viable for
broad cross-national analysis.
116
Appendix Figure 5.1 Patterns of UNITY and RICE scores, and effects on
weighted versus unweighted indices, as CLOSEness varies.
117
Counting Votes
or if
PREFij ¼ Reject, AND Outcomej ¼ Approve, AND TotalAYEj
AYEij – NONVOTESij < Thresholdj
where, for every vote, j:
Thresholdj ¼ number of votes necessary to approve the measure
Outcomej ¼ [Approve, Reject]
RLOSERij and ULOSERij identify votes on which, given how all other
parties voted, a party lost despite the fact that it could have prevailed had it
been fully unified or mobilized. Appendix Table 5.2 illustrates some sce-
narios in a hypothetical legislature with 100 members and two parties, A
and B, with 60 and 40 seats, respectively.
A party ‘‘loses’’ a vote whenever the outcome runs contrary to that
supported by a majority of its voting members. When the parties vote
together, neither loses, as in Votes 1 and 2. When both parties vote along
party lines (high RICEij scores for both), as in Votes 3 and 4, Party B loses
provided that UNITYij scores are closely correlated – that is, provided that
both parties mobilize around the same proportion of their members to
vote. Note, however, that on Vote 4, Party B could have prevailed, given
how Party A voted, had it mobilized its full complement of legislators to
vote nay. Thus, on Vote 4, Party B is a ULOSER.
Votes 5 through 8 represent losses by Party A because of disunity of
various sorts. Vote 5 shows a straightforward breakdown within Party A,
with some members voting against the party majority, swinging the out-
come in favor of the united and mobilized Party B. The results of Vote 6
are similar, where Party A suffers a combination of defections among
voting and nonvoting members while confronting a unified and mobi-
lized opposition. In Votes 7 and 8, Party A is both divided and fails to
mobilize, while Party B lags on either one or the other count but prevails
on the vote outcome. In each of these cases, Party A’s disunity allows B to
win (i.e., costs A the vote). On Votes 5, 6, and 7, Party A could have won
(i.e., passed the measure) if all of its voting members had voted aye, and
obviously had it fully mobilized. In these cases, therefore, Party A is both
the RLOSER and ULOSER. On Vote 8, Party A could have won had it
fully mobilized, but not merely had all its voting members voted aye, so it
is a ULOSER but not an RLOSER.
As long as UNITYij is strongly correlated across the parties on a given
vote, the outcome will reflect the distribution of seats across parties.
118
Appendix Table 5.2. Illustrations of RLOSER and ULOSER in a Hypothetical 100-Member Legislature (tallies are [aye-nay-nonvote])
Vote TALLYAj TALLYBj Losing RICEAj RICEBj RLOSER UNITYAj UNITYBj ULOSER
Party
1 [60–0–0] [40–0–0] None 1.00 1.00 None 1.00 1.00 None
2 [30–0–30] [20–0–20] None 1.00 1.00 None .50 .50 None
3 [60–0–0] [0–40–0] B 1.00 1.00 None 1.00 1.00 None
4 [30–0–30] [0–20–20] B 1.00 1.00 None .50 .50 Party B
5 [45–15–0] [0–40–0] A .50 1.00 Party A .50 1.00 Party A
6 [40–10–10] [0–40–0] A .60 1.00 Party A .50 1.00 Party A
7 [20–10–30] [0–20–20] A .33 1.00 Party A .17 .50 Party A
8 [15–5–40] [10–30–0] A .50 .50 None .17 .50 Party A
119
Counting Votes
120
Appendix 5
treat them as such – that is, to count nonvotes as nays when their effects on
outcomes are equivalent to nay votes.
This approach warrants careful consideration, however, because count-
ing nonvotes as nays renders RICEij scores, in particular, and the indices
built from them, highly sensitive to nonvoting. For example, Appendix
Figure 5.2 compares WRICEi indices for parties in the Nicaraguan
121
Counting Votes
8
The columns do not necessarily reach the 100 percent mark, as indices are not calculated for
independents who remained unaffiliated with any party bloc or for single-member parties.
122
Appendix 6
123
Counting Votes
124
6
125
Explaining Voting Unity
126
Competing Principals and Existing Accounts of Party Unity
127
Explaining Voting Unity
from the chamber in others, and shared with independently elected exec-
utives in others.
In short, the data and theory presented here provide substantial leverage
in testing for party unity driven by cohesiveness and discipline across the
full spectrum of legislative votes, but more limited leverage in testing for
unity driven by the specific mechanism and conditions posited by agenda
cartel theory. The results here shed light on cohesiveness of preferences
among copartisans and the relative monopoly on discipline imposed by
legislative party leaders.
128
Weighted and Unweighted Indices
consensus at the assembly level does not. The more hotly votes are contested
in the legislature overall, the more ‘‘room’’ there is, arithmetically, for dis-
unity within parties. Thus, in the discipline-free scenario, RICEij scores
decline as CLOSEj rises, and WRICEi is lower than URICEi.
Now consider the scenario with party discipline – that is, where party
leaders are able to compel their legislators to vote together. RICEij scores
must still be high on consensual votes, by definition. Where votes are
moderately contested, there is the potential for disunity within parties.
But as votes approach toss-ups (i.e., as CLOSEj approaches 1.0), such that
prevote head counts suggest a handful of switched votes one way or the
other could turn the outcome, party leaders should be increasingly inclined
to impose discipline (King and Zeckhauser 2003). Thus, the more that
discipline, as opposed to cohesiveness, accounts for levels of unity, the
more we should observe elevated RICEij scores as CLOSEj approaches
1.0 – that is, on the votes that enter most heavily in my weighting scheme.
It follows that the more that discipline, as opposed to cohesiveness, drives
voting unity, the higher the ratio of WRICEi to URICEi.1
In Table 6.1 and Figure 6.1, Party A illustrates the discipline-free sce-
nario, and Party B the disciplined scenario, for RICE scores and indices
across six votes in a hypothetical 300-member legislature. In Party A,
RICEAj declines as successive votes are contested more closely at the level
of the legislature as a whole; thus, the weighted index is substantially lower
than the unweighted. Party B experiences divisions on some moderately
contested votes but pulls together on the closest votes, with the effect that
its weighted index is higher than its unweighted. Party C experiences only
one instance of dissent, on a close vote, such that its weighted index, like
A’s, is lower than its unweighted (although not by as much).
A relevant statistic, then, is the ratio between WRICEi and URICEi.2 A
lower ratio indicates a tendency for intraparty splits, when they do occur, to
1
Snyder and Groseclose (2000) use a variation on this insight to gain leverage on the
cohesiveness-versus-discipline debate on roll call voting in the U.S. Congress. They suggest
that, whatever levels of dissent are evident, lopsided votes should nevertheless contain in-
formation about legislators’ unconstrained policy preferences (and so, about cohesiveness),
on the grounds that party leaders should have no interest in imposing discipline on votes
that are not expected to be close.
2
Note that the information in the analogous relationship between WUNITYi and
UUNITYi is, again, more ambiguous. UNITYij can be low if many members abstain or
do not vote, even on consensual votes (i.e., low CLOSEj). As a result, the relationship
between weighted and unweighted RICEi indices provides better purchase than that be-
tween weighted and unweighted UNITYi indices on cohesiveness versus discipline.
129
Explaining Voting Unity
Table 6.1. Cohesiveness, Discipline, and Weighted and Unweighted RICE Indices in
a Hypothetical Legislature
Figure 6.1 RICE scores for three parties: Cohesiveness versus discipline.
happen on closely contested votes. Higher ratios indicate parties that may
experience splits on lopsided votes but pull together on closer ones. Across
the cases examined here, it turns out that weighted indices tend to be
slightly lower than unweighted, but there is substantial variance. The mean
WRICE:URICE ratio ¼ .95, but the standard deviation ¼ .12. So in the
typical party, unity does not deteriorate ‘‘when it matters most’’ as
130
Weighted and Unweighted Indices
chronically as for Party A, but neither does the average party rally when the
chips are down as reliably as Party B.
If we assume that party leaders value unity more, and are less tolerant of
cross-voting, on close votes than lopsided ones, then the ratio is an indicator
of the leader’s ability to impose discipline (Snyder and Groseclose 2000;
King and Zeckhauser 2003). The ratio, therefore, provides a rough proxy
for the relative contribution of discipline to a party’s overall voting unity.
It is worth noting that ratios do not necessarily correspond to overall
levels of party unity. The following combinations are all possible:
low unity, low ratio, indicating a party that is neither cohesive nor
disciplined;
low unity, high ratio, indicating a pervasive lack of cohesiveness, but the
party pulls together more on close votes than on lopsided ones,
suggesting some measure of discipline;
high unity, low ratio, indicating a party that is generally cohesive, but
what breaches occur come on close votes, suggesting a lack of disci-
pline; or
high unity, high ratio, indicating a party that is consistently unified, and
airtight on close votes.
Comparing a party’s ratio with its level of voting unity can inform us
about the relative contributions of cohesiveness and discipline to overall
unity. In making this comparison, it is preferable to use URICEi rather
than WRICEi, the formula for which emphasizes close votes and so already
encompasses more of the information reflected in the ratio itself. Consider
Figure 6.2.3
There are cases of discipline without unity (i.e., low unity, high ratio) in
the lower right section, but the unity-without-discipline upper left corner of
the graph is sparsely populated. The ratio and URICEi are correlated at .34,
and the general pattern is that parties with higher ratios are more unified.4
3
The Philippines case is dropped because it is an extreme outlier in its near absence of even
moderately contested floor votes, leaving ratios that are hypersensitive to a handful of close
votes (see Table 5.1).
4
Plotting WRICEi against the weighted:unweighted ratio yields a tighter scatter, with no
low-ratio parties extremely low on WRICEi and a stronger positive correlation (.63), al-
though the lower right quadrant (high discipline, low unity) remains well populated. This
stronger relationship is to be expected, given that both WRICEi and the ratio are drawing
on the same proclivity of copartisans to vote together on close votes. Nevertheless, the
pattern confirms that among highly unified parties, the proclivity to vote with one’s co-
partisans is more pronounced on close votes, precisely when party leaders should be watching.
131
Explaining Voting Unity
The relationship is far from ironclad, and there are cases of all the combi-
nations of cohesiveness and unity outlined previously, but the overall
pattern is consistent with the idea that some measure of discipline on
close votes, in addition to innate cohesiveness, accounts for party unity
in legislative voting.
132
Legislative Parties and Competing Principals
Thus, virtually all legislators are subject to influence by at least one prin-
cipal – their legislative party leadership.
Whether they are subject to pressure from other, competing principals
depends on the institutional context in which they operate. The hypotheses
that follow posit the impact of various factors that affect the relative com-
mitment of legislators to the national party’s collective reputation and
those that could pull legislators in other directions.
5
The degree to which electoral rules encourage personal vote seeking among candidates can
be parsed more precisely than the dichotomy on which I rely here. Even in Carey and
Shugart’s (1995) full rank ordering of electoral systems, however, the key distinction –
which determines whether increases in district magnitude will push toward more or less
personalism – is whether or not voters are afforded the opportunity to cast votes among
competing copartisans that determine which candidates within the party win seats. Given
the amount of empirical variation on electoral rules represented in my data, the best option
is to rely on this yes-no dichotomy on intraparty competition to characterize institutional
forces operating on the connection between legislators and their principals.
133
Explaining Voting Unity
H2: Party unity should be lower in federal systems than in unitary ones.
134
Legislative Parties and Competing Principals
1999; Figueiredo and Limongi 2000; Aleman and Tsebelis 2005). Presi-
dents also command the attention of national media, and Calvo (2007)
demonstrates that strong public approval enhances presidents’ ability to
sway legislators toward their proposals.
Given their considerable resources, where presidents’ demands contra-
dict those of legislative party leaders, their influence should undermine
voting unity in systems with popularly elected presidents relative to pure
parliamentary systems. To the extent that presidents might use their
resources to pull votes from any and all legislators, the effect should be
present across all parties.
H3: Party unity should be lower in systems with popularly elected pres-
idents than in pure parliamentary system.
Next consider the difference between regimes with presidents and those
without with respect to how a party’s status, in government or in opposi-
tion, ought to affect party unity. The resources of the executive branch
reinforce the influence of legislative party leaders in the absence of a pres-
ident, but they can undermine this influence if vested in an independent
president. Consider first the no-president scenario. In parliamentary sys-
tems, the party leadership is the principal most influential over any given
legislator, and in the case of government parties, the legislative party lead-
ers and the executive are one and the same. Where legislative leadership
and executive leadership are fused, parties in government have more
resources to impose discipline than do those in opposition (Laver and
Shepsle 1996). This suggests:
H3a: Party unity should be higher in governing parties than opposition
parties under parliamentarism.
Under presidentialism, and in hybrid regimes with elected presidents,
the situation is more complex. The general story from Hypothesis 3, that
presidents compete with party leaders for legislator loyalties, applies, but
presidential attention and influence fall more heavily on the presidents’
copartisans and coalition partners than on legislators from outside the
presidents’ coalition. During interviews in various presidential countries,
I asked, ‘‘On what are voting coalitions based – common ideology, party,
electoral interests, control of the legislative agenda, support for the exec-
utive, etc.?’’ Party was the most commonly cited foundation for legislative
voting, consistent with the premise that legislators are beholden to legis-
lative party leaders. Next most frequent, however, were responses that
135
Explaining Voting Unity
When a party wins, that party group generally forms a stronger connection to the
executive. The strongest relationship is legislative group-to-executive – president of
the Republic, ministers and all the apparatus of public administration. The losing
party group does not maintain much of a strong connection with its party organi-
zation. Parties in this country are not strong ideological structures, such as would
elicit discipline from each deputy. Parties at the national level have been converted
into electoral platforms more than the classical concept of an ideological bloc.
The executive normally works better when it has an assembly majority. The
majority party tries to support projects from the executive, of which it is part. . . .
So when the executive wants to submit a law, it calls on the majority party, explains
the benefits of approving the law, and generally we vote in line with the directives we
are given. This doesn’t mean we are obliged to, but normally that’s what happens.
The Liberal group has been very obedient, through presidential discipline more
than party discipline. They take almost no decisions autonomously from the pres-
ident, and when they have done so, they have had to backtrack when it produces
a presidential veto. One or another deputy has voted against the president’s wishes,
and then along comes some bit of patronage that makes him change his vote, and we
vote again the way the president orders. [Interviewer: What are examples of patron-
age?] Public jobs, for deputies and relatives. The deputy might be made ambassador
to some country, and maybe they send his family or relatives. . . . The rest of the
deputies, that are not from the Sandinista or Liberal groups, many of them have
formed alliances with the government . . . [but] these aren’t real political alliances,
but rather alliances based on patronage.
136
Legislative Parties and Competing Principals
6
For example, Weldon (1997) demonstrates that the source of unity in the archetypal case of
airtight party discipline under presidentialism – Mexico’s Partido Revolucionario Institu-
cional during its long hegemony – was the elaborate institutional structure that afforded
presidents not only their constitutional authorities but also control over all the partisan and
procedural resources that, in other political systems, normally fall in the domain of legis-
lative party leaders. The secret of PRI unity throughout much of the twentieth century was
that legislators had only one meaningful principal.
137
Explaining Voting Unity
138
Legislative Parties and Competing Principals
139
Explaining Voting Unity
140
Picturing Party Unity across Systems
141
Explaining Voting Unity
2005).7 Figure 6.3 presents WRICEi indices for the parties in each country
for which I have data according to whether the constitution includes a con-
fidence vote provision and whether assembly elections provide for compe-
tition among candidates from the same party. In the bottom left panel are
systems with the confidence vote and without intraparty competition. By
and large, voting unity as measured by WRICEi is high, with the average
more than .90. Parties in France’s Fourth Republic are widely regarded to
have been chronically factionalized, but even its mean WRICEi is .85.
Canada and Israel each have a derelict outlier, but in each case these are
two-member parties in which a 1–1 split vote would drive the RICEij score
to zero. Overwhelmingly, the legislators in these parliamentary systems
voted together with their copartisans.
The bottom right panel shows the one case of a confidence vote system
with intraparty competition, Poland. Note that, in addition to the confidence
7
Federalism is frequently mentioned as a source of party disunity at the national level, but in
the most careful analysis of the issue to date, Desposato (2004b) finds evidence for only
a remarkably small drag on voting unity in Brazil, where low party unity had often been
attributed to a federalism effect (Weyland 1996; Mainwaring 1999). Jones and Hwang
(2005) similarly are unable to detect an effect of provincial forces on voting in the Argentine
Congress, although they confront substantial challenges in identifying what effect alliance
with governors ought to have on allied deputies.
142
Picturing Party Unity across Systems
8
Attentive readers might note that both the Peruvian (Art. 134) and Russian (Art. 111)
constitutions provide for removal of the cabinet by vote of a parliamentary majority. How-
ever, both constitutions also allow the president to dissolve the legislature in this instance,
raising the costs to legislators enormously of wielding the no-confidence vote over presi-
dential resistance. Given these provisions, I do not code Peru or Russia as no-confidence
vote systems.
143
Explaining Voting Unity
144
Picturing Party Unity across Systems
145
Explaining Voting Unity
without. (There are no recorded vote data from parliamentary systems with
intraparty competition.) There is no analogous pattern among WUNITYi
indices, which exhibit wider spreads and no apparent responsiveness to
regime type or electoral rules. Figure 6.6, with the darker boxes represent-
ing RLOSERi and the lighter ULOSERi, shows a similar contrast, with
larger increasing RLOSERi values from parliamentary to presidential sys-
tems, and from no intraparty competition to intraparty competition within
presidential systems, but no corresponding pattern for ULOSERi. On the
whole, the mobilization-based indices, WUNITYi and ULOSERi, appear
to be more susceptible than the cross-voting-based indices, WRICEi and
RLOSERi, to distortion via nonvoting equilibria, as discussed in Chapter 5,
which limits their usefulness for cross-national comparisons.
On the whole, the graphical patterns of WRICEi and RLOSERi suggest
that party unity is lower in presidential regimes and in the presence of intra-
party electoral competition. The statistical analysis below adds variables, op-
erating at both the level of political system and the level of individual parties,
to gain additional purchase on sources of party voting unity, and disunity.
6.6. Models
146
Models
9
I thank an anonymous Cambridge University Press reviewer for pointing me toward the
Brambor, Clark, and Golder (2006) article, which illustrates the importance of including all
component variables along with any interactive term. An earlier version of this study (Carey
2007), not fully appreciating this imperative, presents models including Confidence Vote,
rather than Presidential, alongside the Government Party * Presidential interactive term.
10
Analyses on the mobilization-based indices, WUNITYi and ULOSERi, are not included
in the analyses presented in this chapter because the sensitivity of those indices to non-
voting makes them less reliable measures of unity, at least until effective means of identi-
fying and accounting for nonvoting equilibria can be established.
147
Explaining Voting Unity
11
Note that both the Effective Number of Parties and the previous system-level variable,
Regime Age, can vary across assemblies within a given legislature. From Table 5.1, for
example, note that the data include votes from three different Brazilian assemblies, two
from Chile, but one from Australia. Potentially, then, there are three levels of data – party
within an assembly, the specific iteration of the assembly, and the constitutional system in
which the iterations of that assembly exist. In the interest of simplicity, however, I combine
these last two levels, assigning mean values for the Effective Number of Parties and Regime
Age across observations from a given legislature. In practice, there is little variance on
either variable across assemblies within each system for the data included.
148
Models
Seat Share is the proportion of seats held by the party in a given assembly.
Government Party is coded 1 if the party holds at least one cabinet port-
folio in the current cabinet; 0 otherwise.
Government Party * Presidential is the interaction of Government Party
with Presidential.
The logic of the independent variables and expectations about their
effects are mostly straightforward from the hypotheses section, but a few
comments are in order. The default status implied by the models is for an
opposition party in a parliamentary system. The variable Presidential esti-
mates the effect on the dependent variable of being in a presidential regime.
The coefficient on Government Party represents the effect of being in
government, as opposed to out, in parliamentary systems. The coefficient
on Government Party * Presidential picks up the marginal difference be-
tween government parties in systems with directly elected presidents and
those in parliamentary systems. The comparison between government
and opposition parties within presidential systems is also of interest, so
below the list of coefficients from each model the tables show the linear
combination of Government Party + Government Party * Presidential –
Presidential.
Seat Share is included as a control variable, but its logic depends on
the dependent variable. When the dependent variable is %WONi, expec-
tations regarding Seat Share are clear-cut – a greater share of seats should
lead to more wins. When the dependent variable is WRICEi, expectations
are less firm. Parties that comprise larger shares of their chambers may
encompass more-diverse viewpoints and thus be subject to disunity. On the
other hand, increasing seat shares generally provides increasing access to the
legislative resources that party leaders employ to elicit compliance and to
mobilize their rank and file (Hurtado interview). Finally, when the depen-
dent variable is the RLOSERi index of vote losses due to disunity, the effect
of Seat Share should be positive, notwithstanding the fact that bigger par-
ties win more, because a split within a larger party should be more likely to
reverse a vote outcome than the same split in a smaller one.12
12
I also ran the models on vote loss due to disunity controlling for WIN%, on the grounds
that only parties that win votes stand to lose some through breakdowns in unity. That is, if
a party’s winning percentage is zero or close to it, we might reasonably expect that it is
merely in perpetual and futile opposition, rather than that it might have won, say, 3 percent
of those lost votes but for internal splits. This turns out not to be the case, however; the
coefficient on WIN% was never close to significant.
149
Explaining Voting Unity
6.7. Results
Models 1a–c, in Table 6.4, regress WRICEi on sets of both system-level
and party-level explanatory variables. Model 1a includes the full set of
variables implied by the preceding hypotheses, plus a control for each
party’s share of assembly seats. It is, perhaps, most straightforward to note
at the outset the hypotheses for which there is no support in any model.
Contrary to H2, there is no evidence that federalism undermines party
voting unity, as measured either by WRICE or by one of the vote-outcome
(%WON or RLOSER) indices. Nor is there evidence that the existence
of a confidence vote provision (H4), regime age (H5), or party system
fragmentation (H7) matters under any model specifications. With regard
to the confidence vote, statistical results are not shown in Table 6.4 because
Confidence Vote cannot be included in the same model with Presidential,
but alternative specifications dropping Presidential (and Government
Party * Presidential) show no measurable effect of Confidence Vote on
any of the dependent variables.
150
Table 6.4. Multilevel Analysis of Legislative Voting Unity within Parties
Dependent Variables
Independent Variables 1a 1b 1c 2 3a 3b 3c
System level
Intraparty Competition .04 (.10) .05 (.10) .16** (.07) .09 (.07) .017* (.010) .018* (.009) .021** (.010)
Presidential .17 (.11) .19* (.10) .06 (.07) .14* (.08) .004 (.010) .002 (.010) .003 (.011)
Federal .02 (.09) .12** (.06) .004 (.009)
Regime Age (log) .02 (.04) .02 (.03) .003 (.004)
Effective Number Parties .01 (.03) .01 (.02) 000 (.003)
Party level
Party Age (log) .02*** (.01) .02** (.01) .02** (.01)
Seat Share .05 (.08) .05 (.08) .01 (.07) .38*** (.11) .035*** (.008) .036*** (.008) .036*** (.008)
Government Party .05 (.04) .05 (.04) .05 (.03) .25*** (.05) .001 (.003) .001 (.003) .001 (.003)
GovPty * Presidential .01 (.05) .01 (.05) .00 (.04) .18*** (.06) .010** (.004) .010** (.004) .012*** (.004)
Constant .77*** (.21) .89*** (.08) .90*** (.05) .51*** (.15) .010 (.020) .001 (.007) .001 (.007)
H0: Government Party + GovPty * .13 (.11) .15 (.11) .01 (.08) .07 (.09) .007 (.011) .009 (.010) .015 (.011)
Presidential – Presidential ¼ 0
Proportion of variance explained
System level .31 .28 .60 .39 .47 .45 .52
Party level .05 .05 .06 .19 .22 .22 .24
N (parties) 206 206 184 309 254 254 229
N (assemblies) 19 19 16 19 19 19 16
Note: Models 1c and 3c exclude observations from Russia, Nicaragua, and Guatemala, where nonvotes are counted as nay votes. * Significant at > .10
151
level. ** Significant at > .05 level. *** Significant at > .01 level. All models include only parties with two or more legislators.
Explaining Voting Unity
Therefore, Models 1b and 1c (as well as 3b and 3c) drop those system-level
variables for which there is no hint of a measurable effect on party unity,
retaining those for which the graphical representation of data, as well as
bivariate analyses, suggests some impact. Model 1b replicates 1a, but with-
out the Federal, Regime Age, and Effective Number of Parties variables.
The estimated impact of presidentialism on voting unity increases slightly,
and the standard error shrinks, such that the coefficient is now significant at
.06. The shift from a parliamentary regime to one with an elected president
drops WRICEi by .18 – nearly a full standard deviation. This suggests
confirmation of H3, yet sorting out the relative impact of presidentialism
and intraparty electoral competition remains thorny.
Note that among these data, all cases of intraparty competition are in
regimes with elected presidents. Recall further, from Chapter 5 (Section
5.3.2 and Appendix 5), that applying the WRICEi index to assemblies where
all nonvotes are counted as nays is particularly awkward and may yield
unduly low values. All three such assemblies in the data, those of Nicaragua,
Guatemala, and Russia, are in presidential systems without intraparty com-
petition (see Figure 6.3). Model 1c drops these observations, with a result-
ing shift in the relative scope, and the statistical significance, of the
Presidential and Intraparty Competition variables. With the nonvote ¼
nay-vote observations excluded, Intraparty Competition is associated with
a .16 drop in WRICEi, significant at .02. The estimated effect of presiden-
tialism remains negative but falls short of significance (analogous to Intra-
party Competition when the nonvote ¼ nay-vote assemblies are included).
In short, presidentialism and intraparty electoral competition both appear
to diminish WRICEi, but although the cross-national breadth of these data
is unprecedented, statistical leverage at the system level is limited, so un-
certainty as to the relative impact of these factors remains.
152
Results
13
In analyses of %WON and RLOSER, the Party Age variable is never significant. Its in-
clusion or exclusion does not affect the direction or significance of any other variables, but
because I was unable to determine Party Age for almost a third of all parties included in the
data, including the variable in any analysis ‘‘costs’’ a lot of observations. Therefore, Party
Age is dropped from Models 2 and 3. The system-level variables are included as controls in
Model 2, although they are not of substantive interest. The effects of party-level variables
are the same whether the system-level variables are included or not.
153
Explaining Voting Unity
14
The coefficient on the linear combination of regime type and governing status falls short of
significance in the models shown in Table 6.4, but is positive and significant at .01 in the
ordered logit version of Model 3c.
154
Results
Figure 6.7 WRICEi, WUNITYi, RLOSERi, and ULOSERi indices for Canada,
1994–97.
15
Indices from Cardoso’s first administration were shown in Figure 5.8, and the picture is
strikingly similar.
155
Explaining Voting Unity
156
Results
157
Explaining Voting Unity
158
Extending the Analysis
159
Explaining Voting Unity
variables describing political institutions and regime age are the same. I do
not include a measure of party system fragmentation, given that the analysis
includes only a single government and opposition coalition within each
assembly. I do, however, include a coalition-level variable, Multiparty
Coalition, indicating whether a coalition consisted of legislators from more
than one party. The Seat Share variable is the percentage of assembly seats
held by all the parties in a given coalition.
As in the b and c models from Table 6.4, I am parsimonious about
including system-level variables in the models on WRICEi and
RLOSERi to preserve degrees of freedom at that level. I include Intra-
party Competition, which is associated with lower WRICEi indices,
and Presidential, given the prominence of the cross-level interactive
variable, Government Coalition * Presidential (Brambor, Clarke, and
Golder 2006).
160
Extending the Analysis
Table 6.6. Multilevel Analysis of Legislative Voting Unity within Governing and
Opposition Coalitions
votes won, and a 5.1 percent reduction in the rate of losses due to cross-
voting – the mean rate of which across all observations is 6.9 percent.
Whereas governing coalitions in parliamentary systems are more unified,
win more votes, and lose less frequently because of disunity than opposition
coalitions, the coefficients on Governing Coalition * Presidential indicate
that this unity boost is diminished in presidential regimes (more cross-
voting, more losses, and more defection losses), although the marginal effect
reaches statistical significance only with regard to winning percentage. The
linear combination of Government Coalition + GovCoal * Presidential –
Presidential (at the bottom of Table 6.6) indicates that governing coalitions
161
Explaining Voting Unity
162
Conclusion
from voting rules alone was premature (Figueiredo and Limongi 2000).
The results here should dispel uncertainty on this count.
163
Explaining Voting Unity
164
7
165
The Individual-Collective Balance
166
Reviewing the Major Points
reflecting only what the politicians think citizens want to hear. Even if so,
though, this would indicate that politicians believe that citizens want
more individualistic and less party-centered representation from their
legislators. What limited evidence is available from public opinion sur-
veys suggests that this is, indeed, what citizens want. At any rate, there
appears to be a disjuncture between what much of the academic literature
on legislative representation prescribes and what politicians and reform-
ers aim to deliver.
Visible votes are an essential component of individual accountability.
They are in scarce supply in many legislatures, but time and technol-
ogy push toward more visible voting.
The mechanics of voting in legislative assemblies can produce a funda-
mental asymmetry in the ability of legislative insiders (most prominently,
legislative party leaders) and outsiders (everyone else, including citizens,
organized interest groups, the media, and academics) to monitor legisla-
tors’ behavior. The historical and institutional contexts differed consider-
ably, but the adoption and expansion of recorded voting in the United
States and in some countries in Latin America have pushed accountability
in similar directions. On the whole, visible voting has been favored by
opposition legislators and has been used to force unpopular measures
advanced by majority parties and coalitions onto the public record. Inter-
views support the idea that recording and publishing legislative votes
facilitate external monitoring, foster fair play in legislative procedures,
discourage legislators from obfuscating their records, allow voters to re-
ward or punish legislators for votes in elections, and may affect legislative
decisions in anticipation of such effects.
The supply of recorded votes is limited in most Latin American legis-
latures, largely because of reluctance about visible voting on the part of
party leaders who strive to maintain a tight grip on the rules of legislative
procedure. Nevertheless, the increasing availability of secure and efficient
electronic voting systems has driven down dramatically the procedural
costs of recording votes and making them visible. Moreover, once some
set of conditions – a presidential initiative, for example, or a successful
minority-led reform – allows for the establishment of visible voting as stan-
dard operating procedure, the practice appears difficult to revoke, perhaps
owing to the widespread belief among politicians that citizens want
more transparency in legislative institutions, even at the cost of strong
party control.
167
The Individual-Collective Balance
168
The Optimal Mix?
169
The Individual-Collective Balance
under a party label and controls resources legislators care about, divergence
in the demands of these principals will reduce legislative party unity. The
case for collective accountability regards party voting unity as a necessary
condition. I have suggested that institutional arrangements that increase
legislators’ responsiveness to principals other than national party leaders
can push in the direction of individual accountability. The tension between
individual and collective accountability raises the inevitable question of
whether there is some optimal mix of the two, and whether the design of
political institutions can affect whether legislative representation hits
that target.
Academic research on accountability up to now does not provide a
conclusive answer. Some of the most creative recent theoretical work on
accountability focuses on governments or representatives as monolithic
selectors of policy (Manin, Przeworski, and Stokes 1999; Fearon 1999;
Ferejohn 1999), or else on the accountability of presidents alone (Stokes
2001). Among research that discusses legislators and parties explicitly, the
enduring argument for strong-party government holds unity as an unqual-
ified collective good, both in providing coherent options to voters in elec-
tions and in delivering decisive government (American Political Science
Association 1950).
The theme remains central in contemporary scholarship. Powell and
Whitten (1993) include party cohesion as one of the four factors that de-
termine whether voters can assign responsibility to their elected represen-
tatives for policy outcomes, and so hold them accountable. Johnson and
Crisp (2003) demonstrate that the ideological predisposition of legislative
majorities can account for economic policies where president-centered
explanations cannot, but that the connection between legislative party plat-
forms and the policies implemented is stronger when electoral rules dis-
courage individualism. The implication is that collective representation
strengthens the connection between what voters ask for and what they
get. Gerring, Thacker, and Moreno (2005) make a more sweeping claim,
that political institutions that centralize government authority in strong
national parties produce superior policy outcomes on dimensions ranging
from bureaucratic efficiency to investment security to public health and
education. The clear prescription is that representation works best when
legislators answer directly and unequivocally to their parties as principals.
The related theme that institutions that encourage party disunity
produce political pathologies is also prominent. Golden and Chang
(2001) attribute corruption scandals in campaign finance to the degree
170
The Optimal Mix?
171
The Individual-Collective Balance
172
The Optimal Mix?
1
The outgoing regime’s motives for choosing this system were not benign. It sought to
cushion the anticipated defeat of its civilian politician allies in the ensuing elections. The
unusual conditions by which Chile’s electoral rule was imposed over the objections of its
incipient governing majority mitigate the problem of endogeneity of institutions that is
intractable in much of comparative politics (Przeworski 2007). As a result, Chile presents an
unusually favorable environment for testing the effects of institutions on political behavior.
2
Note that I refer here to electoral alliances as running lists, rather than parties. This is
because Chilean electoral law allows parties to coalesce to run lists, and indeed almost all
seats in Chilean elections under this system have been won by candidates from coalition
lists. As a result, Chilean elections have been characterized by intracoalition competition
but not by intraparty competition.
3
A figure based on votes from the third congress is virtually identical. The mobilization-
based indices, WUNITYi and ULOSER, are suspect in the Chilean case, because the
uniform drop-off across parties from WRICEi to WUNITYi suggests nonvoting equilibria.
173
The Individual-Collective Balance
Democratic (DC) party is shown at the left of each panel in black with the
other parties in the governing coalition in white, and opposition parties in
varying gray hues toward the right of each panel.
The parties in Chile’s governing coalition experienced more cross-
voting (mean WRICEi ¼ .86) than do governing parties in pure parliamen-
tary systems (mean WRICEi ¼ .93), but less than governing parties in other
presidential systems (mean WRICEi ¼ .76). Chile’s opposition parties were,
174
The Optimal Mix?
on average, just slightly more unified (mean WRICEi ¼ .87) than their
government counterparts. Opposition parties experienced no floor losses
due to cross-voting in this period and governing parties were RLOSERs
on 0.6 percent of votes – the same rate as governing parties in parliamentary
regimes, and below the 3 percent average among governing parties else-
where under presidentialism. These figures suggest sufficient party unity so
that collective representation is viable in Chile, even while individual legis-
lators seek personal votes and voters maintain the ability to retain or reject
specific representatives. In the election following the period on which
Figure 7.1 is based, 85 of 120 Chamber incumbents were nominated for
reelection, and of these, 72 – that is, 85 percent of those on the ballot,
60 percent of all incumbents – won a successive term.
Chilean political institutions are not beyond criticism, and the elec-
toral system, in particular, is subject to regular proposals for reform
(Altman 2005; Huneeus 2006).4 Yet Chilean democracy since 1990 has
delivered a respectable combination of collective and individual legisla-
tive accountability (Cox 2006b). Elections have produced governing coa-
litions that are easily identifiable by voters and that, once in office, have
generally advanced the policies and platforms on which they campaigned.
Governments have met with regular, although not uniform, success on
the floor of Congress. Chilean presidents throughout this period have
occasionally been publicly at odds with the leaders of their allied parties
and coalition, but the more common scenario has been mutual cooperation.
As a result, government-sponsored legislative proposals have mostly been
successful, although sometimes only after prolonged periods of legislative
deliberation (Siavelis 2000; Londregan 2001). The conventional economic
indicators by which governments are judged have been consistently strong in
Chile during this period, and voters have rewarded the governing coalition
with reelection to the presidency three times and returned majorities from
that coalition to the legislature in four consecutive elections. These same
4
The most persistent criticisms stem from the lack of proportionality and the high barriers to
entry inherent in the two-member district system. Increasing district magnitude could rem-
edy this, as opportunities for minority-party representation increase rapidly with increments
in magnitude in this range. In order to improve proportionality while maintaining the con-
ditions for effective individual accountability and avoiding rampant individualism, reformers
might retain candidate preference voting but embrace only modest increases in magnitude.
At magnitudes above 5 or 6, the informational demands on Chilean voters imposed by open-
list competition would threaten individual accountability, and the incentives for individual-
ism could undermine party unity sufficiently to threaten collective accountability.
175
The Individual-Collective Balance
voters can select individual legislators from among fields of candidates small
enough that campaigns are not mere cacophonies of individualistic appeals.
They have taken the opportunity to exercise that discretion, rewarding some,
but not all, incumbents with reelection.
176
Appendix: Interview Subjects by Country
(continued)
177
Appendix
(continued)
178
Appendix
179
References
Adserá Alicia, Carles Boix, and Mark Payne. 2003. ‘‘Are You Being Served? Political
Accountability and Quality of Government.’’ Journal of Law, Economics, and
Organization 19(2): 445–90.
Alcántara, Manuel. 1994–2000. Proyecto Elites Latinoamericanos (PELA). Salamanca,
Spain: Instituto Interuniversitario de Iberoamérica (Universidad de Sala-
manca).
Aldrich, John H. 1995. Why Parties? The Origin and Transformation of Political Parties
in America. Chicago: University of Chicago Press.
Aleman, Eduardo, and George Tsebelis. 2005. ‘‘Presidential Conditional Agenda
Setting in Latin America.’’ World Politics 57(3): 396–420.
Altman, David. 2005. ‘‘De un sistema paralelo a un compensatorio (proporcional person-
alizado).’’ Santiago, Chile: Instituto de Ciencia Politica, Pontificia Universidad
Catolica de Chile.
American Political Science Association. 1950. Toward a More Responsible Two-Party
System. New York: Rinehart.
Ames, Barry. 1995. ‘‘Electoral Strategy under Open List Proportional Represen-
tation.’’ American Journal of Political Science 39(2): 406–33.
2002. ‘‘Party Discipline in the Chamber of Deputies.’’ In Legislative Politics in
Latin America, ed. Scott Morgenstern and Benito Nacif, 185–221. Cambridge:
Cambridge University Press.
Amorim Neto, Octavio. 2002. ‘‘Presidential Cabinets, Electoral Cycles, and Coalition
Discipline in Brazil.’’ In Legislative Politics in Latin America, ed. Scott Morgenstern
and Benito Nacif, 48–78. Cambridge: Cambridge University Press.
2006. Presidencialismo e governabilidade nas Americas. Rio de Janeiro: Editora
Fundacion Getulio Vargas.
Amorim Neto, Octavio, Gary W. Cox, and Mathew D. McCubbins. 2003. ‘‘Agenda
Power in Brazil’s Camara dos Deputados, 1989–98.’’ World Politics 55(4):
550–78.
Asamblea Nacional de Panamá. 2005. http://www.asamblea.gob.pa/.
Associación de Derechos Civiles. 2004. ‘‘Ley de Reforma Laboral: Falencias de su
votación.’’ Buenos Aires, Argentina.
181
References
182
References
Carey, John M., and Gina Yanitell Reinhardt. 2004. ‘‘State-Level Institutional
Effects on Legislative Coalition Unity in Brazil.’’ Legislative Studies Quarterly
29(1): 23–47.
Carey, John M., and Matthew Soberg Shugart. 1995. ‘‘Incentives to Cultivate
a Personal Vote: A Rank Ordering of Electoral Formulas.’’ Electoral Studies
14(4): 417–39.
1998. ‘‘Calling Out the Tanks, or Filling Out the Forms?’’ In Executive Decree
Authority, ed. John M. Carey and Matthew Soberg Shugart, 1–32. Cambridge:
Cambridge University Press.
Carroll, Royce, Gary W. Cox, and Monica Pachon. 2006. ‘‘How Political Parties
Create Democracy, Chapter 2.’’ Legislative Studies Quarterly 31(2): 153–74.
Centro de Estudios Públicos. 2007. Estudio nacional de opinion pública. Santiago,
Chile: Centro de Estudios Públicos.
Cevasco Piedra, José. 2000. El Congreso del Perú: Un modelo de modernizacio´n. Lima:
Ediciones del Congreso del Perú.
Cheibub, Jose Antonio, and Adam Przeworski. 1999. ‘‘Democracy, Elections, and
Accountability for Economic Outcomes.’’ In Democracy, Accountability, and
Representation, ed. Bernard Manin, Adam Przeworski, and Susan C. Stokes,
222–50. Cambridge: Cambridge University Press.
Cheibub, José Antonio, Adam Przeworski, and Sebastian M. Saiegh. 2004.
‘‘Government Coalitions and Legislative Effectiveness under Presidentialism
and Parliamentarism.’’ British Journal of Political Science 34: 565–87.
Chhibber, Pradeep, and Ken Kollman. 1998. ‘‘Party Aggregation and the Number
of Parties in India and the United States.’’ American Political Science Review
92(2): 329–42.
Cları´n. 2004a. ‘‘La justicia resolvera si debe quedar registrado como vota cada
diputado porteno.’’ Buenos Aires, Argentina. February 17. www.clarin.com.
2004b. ‘‘Otra forma de votar las leyes.’’ Buenos Aires, Argentina. February 18.
www.clarin.com.
Cleary, Matthew R., and Susan C. Stokes. 2006. Democracy and the Culture of Skep-
ticism: Political Trust in Argentina and Mexico. New York: Russell Sage Foun-
dation.
Colomer, Josep M. 2001. Political Institutions: Democracy and Social Choice. New
York: Oxford University Press.
Condorcet, Marquis de. 1785. Essay on the Application of Analysis to the Probability of
Majority Decisions (Essai sur l’application de l’analyse à la probabilté des deci-
sions rendues a la pluralité des voix). Paris: De l’Imprimerie royale.
Congreso del Perú. 1998. Diario de los debates. Primera Legislatura Ordinaria de
1998, 11a Sesión, 24 de setiembre.
Coppedge, Michael J. 1994. Strong Parties and Lame Ducks: Presidentialism,
Partyarchy, and Factionalism in Venezuela. Stanford: Stanford University Press.
1998. ‘‘The Dynamic Diversity of Latin American Party Systems.’’ Party Politics
4(4): 547–68.
Corrado, Anthony. 2002. ‘‘A History of Federal Campaign Finance Law.’’ In The
New Campaign Finance Sourcebook, ed. Anthony Corrado, Thomas Mann,
183
References
184
References
185
References
Garman, Christopher, Stephan Haggard, and Eliza Willis. 2001. ‘‘Fiscal Decen-
tralization: A Political Theory with Latin American Cases.’’ World Politics
53(2): 205–36.
Gerring, John, and Strom C. Thacker. 2008. A Centripetal Theory of Democratic
Governance. Cambridge: Cambridge University Press.
Gerring, John, Strom C. Thacker, and C. Moreno. 2005. ‘‘Centripetal Democratic
Governance: A Theory and Global Inquiry.’’ American Political Science Review
99(4): 567–81.
Golden, Miriam A., and Eric C.C. Chang. 2001. ‘‘Competitive Corruption:
Factional Conflict and Political Malfeasance in Postwar Italian Christian
Democracy.’’ World Politics 53(4): 588–622. http://muse.jhu.edu/journals/
world_politics/ v053/53.4golden.pdf.
Gonzalez Bertomeu, Juan F. 2004. Cada voto con su nombre: Votaciones nominales en el
Poder Legislativo. Buenos Aires: Asociación por los Derechos Civiles.
Gunson, Phil. 2006. ‘‘Chavez’s Venezuela.’’ Current History 105(688): 58–63.
Hallerberg, Mark, and Patrik Mairer. 2004. ‘‘Executive Authority, the Personal
Vote, and Budget Discipline in Latin American and Caribbean Countries.’’
American Journal of Political Science 48(3): 571–87.
Haspel, Moshe, Thomas Remington, and Steven S. Smith. 1998. ‘‘Electoral Insti-
tutions and Party Cohesion in the Russian Duma.’’ Journal of Politics 60(2):
417–39.
Hellwig, Timothy, and David Samuels. 2007. ‘‘Electoral Accountability and the
Variety of Democratic Regimes.’’ British Journal of Political Science 38: 65–90.
Hix, Simon. 2002. ‘‘Parliamentary Behavior with Two Principals: Preferences,
Parties, and Voting in the European Parliament.’’ American Journal of Political
Science 46(3): 688–98.
2004. ‘‘Electoral Institutions and Legislative Behavior – Explaining Voting
Defection in the European Parliament.’’ World Politics 56(2): 194–223.
Hix, Simon, Abdul Noury, and Gerard Roland. 2006. ‘‘Dimensions of Politics in
the European Parliament.’’ American Journal of Political Science 50(2): 494–511.
Huber, John D. 1996. ‘‘The Vote of Confidence in Parliamentary Democracies.’’
American Political Science Review 90(2): 269–82.
Huber, John D., and G. Bingham Powell. 1994. ‘‘Congruence between Citizens
and Policymakers in Two Visions of Liberal Democracy.’’ World Politics 46:
291–326.
Huneeus, Carlos. ‘‘Binominalismo y calidad democratica.’’ La Tercera, Santiago,
Chile, August 8, 2006, Opinion sec.
International Institute for Democracy and Electoral Assistance. 1997. Handbook of
Electoral Systems. Stockholm, Sweden: IDEA.
Jenkins, Jeffrey A. 1999. ‘‘Examining the Bonding Effects of Party: A Comparative
Analysis of Roll-Call Voting in the US and Confederate Houses.’’ American
Journal of Political Science 43(4): 1144–65.
2000. ‘‘Examining the Robustness of Ideological Voting: Evidence from the
Confederate House of Representatives.’’ American Journal of Political Science
44(4): 811–22.
186
References
Jenkins, Jeffrey A., and Charles Stewart III. 2003. ‘‘Out in the Open: The Emer-
gence of Viva Voce Voting in House Speakership Elections.’’ Legislative Studies
Quarterly 28(4): 481–508.
Johnson, Chris, William Maley, Alexander Thier, and Ali Wardak. 2003. Afghani-
stan’s Political and Constitutional Development. London: Overseas Development
Institute and U.K. Department for International Development.
Johnson, Gregg, and Brian F. Crisp. 2003. ‘‘Mandates, Powers, and Policies.’’
American Journal of Political Science 47(1): 127–41.
Jones, Mark P., and W. Hwang. 2005. ‘‘Party Government in Presidential Democ-
racies: Extending Cartel Theory beyond the US Congress.’’ American Journal
of Political Science 49(2): 267–82.
Katz, Richard S. 1994. ‘‘The 1993 Parliamentary Electoral Reform.’’ In Italian
Politics, ed. Carol Mershon and Gianfranco Pasquino, 93–112. 9th ed. Boulder,
CO: Westview Press.
Kile, Orville M. 1948. The Farm Bureau through Three Decades. Baltimore: Waverly
Press.
King, David C., and Richard J. Zeckhauser. 2003. ‘‘Congressional Vote Options.’’
Legislative Studies Quarterly 28(3): 387–411.
Krehbiel, Keith. 1998. Pivotal Politics: A Theory of U.S. Lawmaking. Chicago:
University of Chicago Press.
2000. ‘‘Party Discipline and Measures of Partisanship.’’ American Journal of
Political Science 44(2): 212–27.
Kulischeck, Michael, and Brian F. Crisp. 2001. ‘‘The Consequences of Electoral
Reform in Venezuela.’’ In Mixed-Member Electoral Systems: The Best of Both
Worlds?, ed. Matthew S. Shugart and Martin P. Wattenberg, 404–31. New
York: Oxford University Press.
Kunicová, Jana, and Susan Rose-Ackerman. 2005. ‘‘Electoral Rules and Constitu-
tional Structures as Constraints on Corruption.’’ British Journal of Political
Science 35(4): 573–606.
La Gaceta. 2001. ‘‘Congresistas que votaron y en contra de la revocatoria.’’ La
Gaceta, 6 May, sec. 2.
La Raja, Raymond J. 2003. ‘‘Why Soft Money Has Strengthened Parties.’’ In Inside
the Campaign Finance Battle: Court Testimony on the New Reforms, ed. Anthony
Corrado, Thomas E. Mann, and Trevor Potter, 69–98. Washington, DC:
Brookings Institution Press.
Laakso, Markku, and Rein Taagepera. 1979. ‘‘Effective Number of Parties: A Mea-
sure with Application to West Europe.’’ Comparative Political Studies 12: 3–27.
Lanfranchi, Prisca, and Ruth Luthi. 1999. ‘‘Cohesion of Party Groups and Inter-
party Conflict in the Swiss Parliament: Roll Call Voting in the National
Council.’’ In Party Discipline and Parliamentary Government, ed. Shaun Bowler,
David M. Farrell, and Richard S. Katz, 99–120. Columbus: Ohio State
University Press.
Latin America Data Base. 2001. ‘‘Nicaragua: Governing Party and Sandinistas
Nominate Presidential Candidates.’’ NotiCen: Central American & Caribbean
Political & Economic Affairs 6(3).
187
References
188
References
McCubbins, Mathew D., and Frances McCall Rosenbluth. 1995. ‘‘Party Provision
for Personal Politics: Dividing the Vote in Japan.’’ In Structure and Policy in
Japan and the United States, ed. Peter F. Cowhey and Mathew D. McCubbins,
35–55. Cambridge: Cambridge University Press.
McKelvey, Richard. 1976. ‘‘Intransitivities in Multidimensional Voting Models:
Some Implications for Agenda Control.’’ Journal of Economic Theory 12: 472–82.
Moe, Terry M., and Michael Caldwell. 1994. ‘‘The Institutional Foundations of
Democratic Government: A Comparison of Presidential and Parliamentary
Systems.’’ Journal of Institutional Theoretical Economics 150(1): 171–95.
Molinar Horcacitas, Juan, and Jeffrey Weldon. 2001. ‘‘Reforming Electoral Sys-
tems in Mexico.’’ In Mixed-Member Electoral Systems: The Best of Both Worlds?,
ed. Matthew Soberg Shugart and Martin P. Wattenberg, 209–30. Oxford:
Oxford University Press.
Morgan, Jana, Rosario Espinal, and Mitchell A. Seligson. 2006. The Political Culture
of the Dominican Republic in 2006. Nashville, TN: Latin American Public Opin-
ion Project.
Morgenstern, Scott. 2003. Patterns of Legislative Politics: Roll Call Voting in Latin
America and the United States. Cambridge: Cambridge University Press.
Morone, James A. 1990. The Democratic Wish: Popular Participation and the Limits of
American Government. New York: Basic Books.
Nicolau, Jairo Marconi. 2007. ‘‘O Sistema Eleitoral De Lista Aberta no Brasil.’’ In
Instituicxo˜es representativas no Brasil: Balancxo e reforma, ed. Jairo Marconi Nicolau
and Timothy J. Power, 97–122. Rio de Janeiro: IUPERJ.
Odegard, Peter H. 1928. Pressure Politics: The Story of the Anti-Saloon League. New
York: Columbia University Press.
Pérez-Liñán, Anı́bal. 2007. Presidential Impeachment and the New Political Instability
in Latin America. Cambridge: Cambridge University Press.
Persson, Torsten, Gerard Roland, and Guido Tabellini. 1997. ‘‘Separation of
Powers and Political Accountability.’’ Quarterly Journal of Economics 112:
1163–1202.
Persson, Torsten, and Guido Tabellini. 2003. The Economic Effects of Constitutions.
Cambridge, MA: MIT Press.
Persson, Torsten, Guido Tabellini, and F. Trebbi. 2003. ‘‘Electoral Rules and
Corruption.’’ Journal of the European Economic Association 1: 958–89.
Pharr, Susan J., Robert D. Putnam, and Russell J. Dalton. 2000. ‘‘A Quarter-
Century of Declining Confidence.’’ Journal of Democracy 11(2): 500–25.
Poole, Keith T. 1998. ‘‘Recovering a Basic Space from a Set of Issue Scales.’’
American Journal of Political Science 42(3): 954–93.
Poole, Keith T., and Howard Rosenthal. 1985. ‘‘A Spatial Model for Legislative
Roll Call Analysis.’’ American Journal of Political Science 29(2): 357–84.
1997. Congress: A Political-Economic History of Roll Call Voting. New York: Oxford
University Press.
2001. ‘‘D-Nominate after 10 Years: A Comparative Update to Congress; A
Political-Economic History of Roll-Call Voting.’’ Legislative Studies Quarterly
26(1): 5–30.
189
References
190
References
Rousseau, Jean Jacques. 1763. ‘‘Voting.’’ In On the Social Contract, book IV, ch. XX.
New York: St. Martin’s Press, 1978.
Rubin, Barnett R. ‘‘The Wrong Voting System.’’ International Herald Tribune,
March 16, International section.
Samuels, David. 2000. ‘‘Ambition and Competition: Explaining Legislative Turn-
over in Brazil.’’ Legislative Studies Quarterly 25(3): 481–97.
2004. ‘‘Presidentialism and Accountability for the Economy in Comparative
Perspective.’’ American Political Science Review 98(3): 425–36.
Samuels, David, and Matthew S. Shugart. 2003. ‘‘Presidentialism, Elections, and
Representation.’’ Journal of Theoretical Politics 15(1): 33–60.
Sartori, Giovanni. 1976. Parties and Party Systems: A Framework for Analysis.
Cambridge: Cambridge University Press.
Scarrow, Susan E. 2001a. ‘‘Direct Democracy and Institutional Change: A
Comparative Investigation.’’ Comparative Political Studies 34(688): 651–65.
2001b. ‘‘Germany: The Mixed-Member System as a Political Compromise.’’
In Mixed-Member Electoral Systems: The Best of Both Worlds?, ed. Matthew
S. Shugart and Martin P. Wattenberg, 55–69. New York: Oxford University
Press.
Seligson, Mitchell A., Abby B. Cordova, Juan Carlos Donoso, Daniel Moreno
Morales, Diana Orces, and Vivian Schwarz Blum. 2007. Democracy Audit:
Bolivia 2006 Report. Nashville, TN: Latin American Public Opinion Project.
Shepsle, Kenneth. 1991. Models of Multiparty Electoral Competition. New York:
Harwood Academic.
Shugart, Matthew S. 1998. ‘‘The Inverse Relationship between Party Strength and
Executive Strength: A Theory of Politicians’ Constitutional Choices.’’ British
Journal of Political Science 28(1): 1–29.
Shugart, Matthew S., and John M. Carey. 1992. Presidents and Assemblies:
Constitutional Design and Electoral Dynamics. Cambridge: Cambridge University
Press.
Shugart, Matthew S., Erika Moreno, and Luis E. Fajardo. 2006. ‘‘Deepening
Democracy by Renovating Political Practices: The Struggle for Electoral
Reform in Colombia.’’ In Peace, Democracy, and Human Rights in Colombia,
ed. Christopher Welna and Gustavo Gallon, 202–66. Notre Dame, IN: Notre
Dame University Press.
Shugart, Matthew S., Melody E. Valdini, and K. Suominen. 2005. ‘‘Looking for
Locals: Voter Information Demands and Personal Vote-Earning Attributes of
Legislators under Proportional Representation.’’ American Journal of Political
Science 49(2): 437–49.
Shugart, Matthew S., and Martin P. Wattenberg, eds. 2001. Mixed-Member Electoral
Systems: The Best of Both Worlds? New York: Oxford University Press.
Siavelis, Peter. 2000. The President and Congress in Postauthoritarian Chile: Institu-
tional Constraints to Democratic Consolidation. University Park: Pennsylvania
State University Press.
Sieberer, Ulrich. 2006. ‘‘Party Unity in Parliamentary Democracies: A Compara-
tive Analysis.’’ Journal of Legislative Studies 12(2): 150–78.
191
References
Singer, Judith D. 1998. ‘‘Using SAS PROC MIXED to Fit Multilevel Models,
Hierarchical Models, and Individual Growth Models.’’ Journal of Educational
and Behavioral Statistics 24(4): 323–55.
Skeen, C. Edward. 1986. ‘‘Vox Populi, Vox Dei: The Compensation Act of 1816
and the Rise of Popular Politics.’’ Journal of the Early Republic 6(253): 274.
Smith, Steven S. 1989. Call to Order: Floor Politics in the House and Senate.
Washington, DC: Brookings Institution Press.
Snyder, James M., and Tim Groseclose. 2000. ‘‘Estimating Party Influence in
Congressional Roll-Call Voting.’’ American Journal of Political Science 44(2):
193–211.
Snyder, James M., and Michael M. Ting. 2005. ‘‘Why Roll Calls? A Model of
Position Taking in Legislative Voting and Elections.’’ Journal of Law, Econom-
ics, and Organization 21(1): 153–78.
Steenbergen, Marco R., and Bradford S. Jones. 2002. ‘‘Modeling Multilevel Data
Structures.’’ American Journal of Political Science 46(1): 218–237.
Stokes, Susan C. 2001. Mandates and Democracy: Neoliberalism by Surprise in Latin
America. Cambridge: Cambridge University Press.
Sundquist, James L. 1981. The Decline and Resurgence of Congress. Washington, DC:
Brookings Institution Press.
Taagepera, Rein, and Matthew Sobert Shugart. 1989. Seats and Votes: The Effects and
Determinants of Electoral Systems. New Haven: Yale University Press.
Tapia G., Sady. 2004. ‘‘Legisladores ‘huyen’ del voto electrónico.’’ La Prensa,
March 25. La Prensa Web, http://www.prensa.com/.
Texas State Legislature. 2005. ‘‘Texas Legislature Online.’’ http://www.capitol
.state.tx.us/tlo/legislation/voteinfo.htm.
Thames, Francis C. 2007. ‘‘Searching for the Electoral Connection: Parliamentary
Party Switching in the Ukrainian Rada, 1998–2002.’’ Legislative Studies Quar-
terly 32(2): 223–56.
Tsebelis, George. 1999. ‘‘Veto Players and Law Production in Parliamentary
Democracies: An Empirical Analysis.’’ American Political Science Review 93(3):
591–608.
Ungar Bleier, Elisabeth. 2002. ‘‘Consideraciones sobre la reforma politica.’’
Testimony before the Comisio´n Primera Constitucional del Senado de la República.
Bogota´.
United States Supreme Court. 1958. Majority Decision in NAACP v. Alabama
Ex Rel. Patterson, Attorney General. No. 91 Supreme Court of the United States
357 U.S. 449; 78 S. Ct. 1163.
Valenzuela, Arturo. 1994. ‘‘Party Politics and the Crisis of Presidentialism in
Chile: A Proposal for a Parliamentary Form of Government.’’ In The Failure
of Presidential Democracy: The Case of Latin America, vol. 2, ed. Juan J. Linz and
Arturo Valenzuela, 91–150. Baltimore: Johns Hopkins University Press.
VoteWorld. 2008. VoteWorld: The International Legislative Roll-Call Voting Website.
http://ucdata.berkeley.edu:7101/new_web/VoteWorld/voteworld/.
Wallack, Jessica S., A. Gaviria, Ugo Panizza, and Eduardo Stein. 2003. ‘‘Particu-
larism around the World.’’ World Bank Economic Review 17(1): 133–43.
192
References
193
Index
accountability, 1–4, 7–9, 11, 16, 29–36, Brazil, 15, 18, 25, 45, 47, 58, 60, 61, 71,
43, 45, 46, 51–55, 63, 64, 65 75, 88, 105, 108, 110, 125, 141,
68–70, 74, 80, 82, 85, 89–91 142, 144, 155, 156, 157, 163
92, 93, 165–172, 175, 176 Burke, Edmund, 69, 70, 91
collective, 1, 2, 9, 10, 12–14, 20–21,
24, 29, 92, 166, 168–176 Calvo, Ernesto, 47, 135
individual, 1–3, 9, 12, 13, 15–16, 20, campaign contributions, 85
21, 32, 36, 41, 44, 166–168 Canada, 103, 104, 105, 107, 108, 141,
170–172 142, 154, 155
rendiciones de cuentas, 33 Carazo Odio, Rodrigo, 79
Afghanistan, 9–13, 172 Cardoso, Fernando Henrique, 110,
agenda powers, 5, 17, 18, 25, 44, 48, 76, 155, 156
77, 78, 81, 91, 125–128, 132 Chavez, Hugo, 27, 32, 33, 34, 90, 137
134–136, 166 Chile, 37, 45, 58, 60, 61, 71, 88, 104,
American Legion, 54 105, 107, 108, 140, 141, 173
American Medical Association, 54 174, 175
American Political Science Association, CLOSE score, 104, 116
4, 23, 170 coalitions, 6, 15, 16, 21, 22, 29, 47, 48,
Ames, Barry, 47, 97, 125 53, 64, 65, 75–77, 78, 80, 88, 90,
Antigone. See Sophocles 92, 93, 99, 109, 110, 113, 126–127,
Anti-Saloon League, 54 135, 155–163, 166–169, 173
Argentina, 15, 45, 50, 58, 59, 60, 61, 64, 174–176
71, 75, 81, 90, 98, 99, 105, 107, governing, 159
108, 140, 141, 142 opposition, 159, 161
Austen-Smith, David, 33 cohesiveness, 7, 27, 123, 126–129
Australia, 103, 105, 107, 108, 109, 141 132, 162
Coleman, William, 52
Bagehot, Walter, 23, 139, 141 Colombia, 25, 26, 27, 38–41, 45, 49, 57,
Baltodano, Mónica, 28, 29, 80, 89, 136 58, 60, 61, 64, 66, 71, 78, 79
Blanco Oropeza, Carlos, 84, 89, 91 80–84, 90
Bolivia, 25, 26, 28, 31, 32, 41, 42, 45, 49, Combellas, Ricardo, 27, 32, 33, 34, 90,
56, 58, 67, 76, 78, 80 137
195
Index
196
Index
197
Index
presidents, 4, 15–18, 24, 39, 40, 74, 82, Samper, Jorge, 28, 86, 87, 136, 137
125, 134, 135, 137, 139, 149, 154, Sanchez de Lozada, Gonzalo, 28, 76, 81
157, 158, 163, 164, 169, 170, 171, secret ballot, 53, 68
175 secret voting, 50, 55
principals, 3, 4, 14–22, 23, 36, 37, 38, separation of powers. See
39, 69, 70, 72, 74, 90, 127, 133, presidentialism
134, 135, 137, 138, 139, 162 signal voting, 50, 70, 72, 74, 77
166–170 single non-transferable vote (SNTV)
proportional representation, 8, 9, 31, elections, 11, 12, 172
143, 168, 172 single-member district (SMD)
closed lists, 8, 10, 41, 125, 133 elections, 9, 13, 44, 71
Proyecto de Elites Latinoamericanas, Sophocles, 20
36 Stokes, Susan, 43, 125, 140, 170
public votes, 44, 45, 46, 49, 50, 51
53, 54, 58, 66, 68, 69, 75, 79, 81 Taiwan, 11
82, 91 Tarek Saab, William, 32, 34, 90
Texas, state legislature, 76, 85, 90
Quinn, Kevin, 46, 47 Toledo, Alejandro (president of Peru),
86, 98
recall elections, 33, 34 transparency, 2, 21, 44, 45, 46, 63, 68,
recorded votes, 21, 22, 36, 46, 47, 51, 69, 74, 78, 80–84, 87, 90, 102, 165,
55, 57, 59, 60, 62, 65, 76, 78, 79 166, 167, 176
80, 81–90, 94, 102, 103, 104 monitoring, 2, 21, 48, 49, 50, 52, 53,
143, 167 54, 59, 70–75, 81, 88, 165, 167
reform, 8, 31 monitoring votes, 21, 50, 54, 55, 56,
regimes, age of, 141, 148 68, 69, 72–75, 83, 167
Republican Party (United States), 24, Web sites, 50, 59, 63, 64, 65, 76, 81
52 Transparency International, 81, 87, 88
responsible party government, 23, 24, U.S. Congress, 6, 7, 47, 49, 51, 52, 53,
91 55, 57, 60, 64, 103, 111, 113, 127,
responsiveness, 1, 3, 8, 29–33, 41, 52, 176
69, 74, 89, 127, 133, 134, 162, 169, Legislative Reorganization Act of
170, 171 1970, 55
rhetorical ideal point estimation, Speaker of the House, 54
46, 47
RICE index, 94, 95, 96, 102, 112, 115, Ukraine, 32
116, 117, 122, 123, 124, 129, 130 ULOSER index, 96, 99, 100, 101, 102,
Rice, Stuart, 94 108, 109, 110, 111, 112, 118, 119,
RLOSER index, 96, 99, 100, 101, 101, 173
102, 108, 109, 110, 111, 112, 118, United States, 21, 24, 43, 44, 45, 49
119, 120, 154 51, 52, 54, 55, 57, 58, 65, 68, 74
roll call voting. See recorded votes 76, 85, 86, 87, 91, 103, 104, 106,
Rose-Ackerman, Susan, 43, 171 107, 108, 110, 112, 116, 141
Rousseau, Jean Jacques, 68, 69, 91 144, 167, 176
Russia, 32, 97, 103, 104, 106, 107, 108, UNITY index, 95, 96, 102, 112, 115,
120, 141, 143 116, 117, 123
198
Index
199
Other Books in the Series (continued from page iii)